Parv-09apache-2.0
phi_so101_8bin_v1 — opening-pause trim table
A companion trimming table for the phi_so101_8bin_v1 dataset that excludes pre-teleop dead air (12.0% of frames) as chunk start points for ACT policy training, without modifying the underlying dataset.
Downloads0
Episodes119
Why This Matters for Physical AI
This trimming table addresses a critical failure mode in imitation learning policies (ACT) where stationary frames during operator dead time create absorbing states that cause the robot arm to freeze during inference, improving training robustness for real-world manipulation tasks.
Technical Profile
- Action Space
- joint_positions
- Task Types
- manipulation
- Episodes
- 119
- Data Format
- csv
- Annotation Types
- action_labels
- License
- apache-2.0
Access
Need custom physical AI data?
Claru builds purpose-built datasets for any environment applications with dense human annotations and quality assurance.
Request a Sample Pack