This dataset was created using LeRobot.
This dataset is derived from k1000dai/so101-write: per-frame task_index was remapped so that the subtask annotations (e.g. "write 3 second stroke") become the dataset tasks, ready for language-conditioned training with LeRobot. Frames outside annotated segments were forward/backward-filled within each episode; iwrite typo labels were fixed.
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30… See the full description on the dataset page:
https://huggingface.co/datasets/k1000dai/so101-write-subtask.
