Add aggregate episode, synchronized-time-step and camera-video totals; clarify the automatic HF video preview; refine source and training wording while preserving tables, task definitions, visual assets and tested code.
Document verified per-task counts and PyTorch training batches
Confirm exactly 100 train episodes per Full-Shot task and one per One-Shot task, add the locally executed DataLoader example and domain-specific model dimensions, and pin downloads to the release with synchronized subset documentation.
Refine RoboValue dataset presentation and locally verified quick start
Document ARX and AgiBot Genie02, reviewed and recollected simulation sources, common trajectory contents, and unified displayed simulation task names. Align bilingual card layout with the RoboValue project README and provide a locally executed LeRobot 0.4.2 example. Training data and provenance remain unchanged.
Clean dataset card into definitive dataset descriptions
Remove release-process narration, minor alignment-row history and provisional wording; retain dataset scope, schemas, task mappings, source attribution and reading instructions. State authorized public release and verified RoboDojo-source licensing. No training data changed.
Present simulation before real-world and complete dataset card guidance
Preserve the overview, domain, layout and download structure; reorder both languages and catalog; document all four released subsets, pinned loading, statistics and One-Shot task correspondence. Training data subtrees remain unchanged.
Document verified Real-World One-Shot training release
Summarize the four subset sizes, use paper task names without platform IDs, keep key state/action information visible, and fold extended field mappings and task descriptions. Training data and videos are unchanged.
8 days ago
Organize bilingual dataset into real-world and simulation subsets
Rename the real-world full-shot training directory, reserve the other three subset directories, document joint and end-effector action fields, and remove operational notes from current dataset documentation.
Label real-world full-shot subset and add bilingual dataset overview
Document the current 20-task, 2000-episode G2A real-world release; distinguish planned one-shot and simulation subsets; preserve all existing training paths and frozen checksums.
