--- pretty_name: RoboDojo Devset Replays tags: - robotics - evaluation configs: - config_name: default data_files: - split: dev path: data/devset.jsonl - config_name: video_action_pairs data_files: - split: dev path: index/video-actions.jsonl --- # RoboDojo Devset Replays Replay data for the 14 methods in the main comparison. Each method uses the same 111 canonical P50_111 cases (42 tasks). **14 × 111 = 1,554 unique method–case records. 4,662 MP4s and 1,554 action files were verified at the artifact revision in [verification.json](verification.json).** [Open the replay explorer](https://huggingface.co/spaces/ZefanW/robodojo-replay-explorer) · [Canonical cases](index/cases.jsonl) · [All replays](data/devset.jsonl) · [Per-method manifests](manifests) · [Video-to-action index](index/video-actions.jsonl) · [Verification report](verification.json) ## Follow the actions in a video The head, left-wrist and right-wrist cameras record the same robot run. All three views share one `actions.json` in their case directory. The video-to-action index has one row per MP4, with its camera name, replay run ID, video path and matching action path. Both files have SHA-256 checksums. For example, open the Compact readings replay for **insert_tubes, standard, seed 0, scene 0**: [head video](replays/compact-readings/insert_tubes__standard__s0__scene0/head.mp4) · [left-wrist video](replays/compact-readings/insert_tubes__standard__s0__scene0/left_wrist.mp4) · [right-wrist video](replays/compact-readings/insert_tubes__standard__s0__scene0/right_wrist.mp4) · [matching actions](replays/compact-readings/insert_tubes__standard__s0__scene0/actions.json). Read `events` in decision order. The `control_start` and `control_end` fields give cumulative robot control steps. When an event supplies `camera_frames` for the selected camera, its frame range is half-open: start inclusive, end exclusive. The index reports how many events have a usable range for that camera. Use the action order and control counts for the remaining events. | Method (per-method manifest) | Cases | MP4s | Actions | Score / 100 | |---|---:|---:|---:|---:| | [Official L3](manifests/official-l3.jsonl) | 111 | 333 | 111 | 29.6562 | | [Full history](manifests/full-history.jsonl) | 111 | 333 | 111 | 34.6771 | | [Adaptive steps](manifests/adaptive-steps.jsonl) | 111 | 333 | 111 | 37.8542 | | [20-step cap](manifests/20-step-cap.jsonl) | 111 | 333 | 111 | 37.9792 | | [Relative targets](manifests/relative-targets.jsonl) | 111 | 333 | 111 | 33.9062 | | [Semantic ops](manifests/semantic-ops.jsonl) | 111 | 333 | 111 | 34.6250 | | [Visual demo](manifests/visual-demo.jsonl) | 111 | 333 | 111 | 39.8646 | | [Text trajectory](manifests/text-trajectory.jsonl) | 111 | 333 | 111 | 35.1771 | | [Self-built library](manifests/self-built-library.jsonl) | 111 | 333 | 111 | 36.9896 | | [Affordances drawn](manifests/affordances-drawn.jsonl) | 111 | 333 | 111 | 36.7396 | | [Compact readings](manifests/compact-readings.jsonl) | 111 | 333 | 111 | 39.8333 | | [Depth, predicted](manifests/depth-predicted.jsonl) | 111 | 333 | 111 | 42.3958 | | [Affordance panel](manifests/affordance-panel.jsonl) | 111 | 333 | 111 | 42.1146 | | [Coordinate solving](manifests/coordinate-solving.jsonl) | 111 | 333 | 111 | 50.3125 | ## Data format Case IDs identify the task, variant, seed and scene. Each `replays///` directory contains `head.mp4`, `left_wrist.mp4`, `right_wrist.mp4` and `actions.json`. The main manifest lists their paths, SHA-256 checksums and byte sizes. Scores use the original 0–1 scale. Each event in `actions.json` includes a nullable `rationale` and `rationale_source`. The text is the public tool call’s original `note`, or its `reason` for giving up. Events without an attributable action note carry null. System budget messages and retrospective comments are excluded. Recorded rationale is available for 97,226 of 98,729 events across 1,553 replays. [Coverage by method](index/rationale-coverage.json). Actions use robodojo.replay.actions.v1. Events distinguish requested targets, observed execution counts and known measured state. Zero-control attempts are preserved. Missing measurements and unavailable frame alignments stay null. Code bodies, private logs and hidden model reasoning are excluded. ## Provenance and limitations - The 13 local methods follow the published [dashboard](https://zefanw.github.io/robodojo-l3-dashboard/) P50_111 selections. Full history includes one explicitly approved supplementary result; its evidence label is retained. - Official L3 scores come from the [original grid](https://github.com/RoboProbe/RoboProbe/blob/3c0bc68383a75bf3fa2d7bfc5ab1d1241c7e171e/results/l3_inspect_eef_official_2100/astra_task_layouts.json). The grid does not give original run IDs. The replays are unique public Astra Console candidates matched by task, variant, actual layout, score and success. The source policy label is RoboDojo_Agent_L4_Inspect_EEF@astra. The replay run ID is separate from score provenance. - Official videos are unchanged copies of the [Solomonz Astra rollouts](https://huggingface.co/datasets/Solomonz/roboprobe-astra-rollouts/tree/61836b7fbde3d6004e43ee0e66a8b99f6e9f4311), whose dataset card states Apache-2.0. See the upstream [Console](https://huggingface.co/spaces/Solomonz/roboprobe-console) and [per-case credits and fixed URLs](provenance/official.jsonl). The upstream public records omit full commanded targets and measured joint poses. - Official store_tools_in_toolbox / standard / scene 0 has conflicting raw-trace success and grid outcome. The official score is preserved and flagged. Official classify_objects_by_language / standard / scene 3 has three late decisions beyond recorded frames. Their frame links are null. ## Check a downloaded copy Run **python validate.py --index-only** to check 14 × 111 identities and per-method manifests. Run **python validate.py** to additionally verify every MP4 and action file against its SHA-256. [Source snapshots](index/source-snapshots.json). ## Paper These replays accompany [JustRUA: A Controlled Study of GPT-6 Robot-Use Agents](https://arxiv.org/abs/2611.16009). [Project page](https://zefanw.github.io/JustRUA/) · [Code](https://github.com/ZefanW/JustRUA) · [Replay explorer](https://huggingface.co/spaces/ZefanW/robodojo-replay-explorer)