Spa-Bench2026apache-2.0
Spa-Bench Evaluation Rollouts — GR00T-N1.7 Full Fine-Tune (Partial)
Physical Spa-Bench rollout trajectories with synchronized middle and wrist RGB video, robot state, action, timestamps, episode indices, and task indices collected at 30 FPS. Covers 120 scored familiar-condition recordings across six task families for evaluating the GR00T-N1.7 model.
Downloads125
Episodes120
Hours1.1773
Why This Matters for Physical AI
This dataset enables benchmark evaluation and reproducibility of the GR00T-N1.7 model's performance on complex spatial reasoning and language grounding tasks in physical robotic manipulation.
Technical Profile
- Modalities
- rgbproprioception
- Robot Embodiments
- so_follower
- Action Space
- 6-DoF: shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, gripper.pos
- Environment
- lab
- Task Types
- countingordinal_positionphysical_statereferential_descriptionrelational_placementrelative_size
- Episodes
- 120
- Total Hours
- 1.1773
- Data Format
- LeRobot
- Annotation Types
- task_labelsepisode_indices
- License
- apache-2.0
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack