a-F1/robotwin-fastwam-lila-clean-exact
RoboTwin Fast-WAM + LiLa-clean exact Exact formal-run data: Fast-WAM head + left/right wrist cameras and synchronized 14-D qpos/action, followed by static-segment cleaning using the LiLa rule at commit 65a320397faae07f311680fcb32f6c2f27dca1a0. This is not the public LiLa-WAM processed package and contains no LiLa code or weights. Source: yuanty/robotwin2.0-fastwam@aac262c35d02cc71b2f6ef670bd65fd9f2bb2547. That fixed source dataset explicitly releases its preprocessed RoboTwin2.0… See the full description on the dataset page: https://huggingface.co/datasets/a-F1/robotwin-fastwam-lila-clean-exact.
RoboTwin Fast-WAM + LiLa-clean exact
Exact formal-run data: Fast-WAM head + left/right wrist cameras and synchronized 14-D qpos/action, followed by static-segment cleaning using the LiLa rule at commit 65a320397faae07f311680fcb32f6c2f27dca1a0. This is not the public LiLa-WAM processed package and contains no LiLa code or weights.
Source: yuanty/robotwin2.0-fastwam@aac262c35d02cc71b2f6ef670bd65fd9f2bb2547. That fixed source dataset explicitly releases its preprocessed RoboTwin2.0 derivative under MIT. The upstream RoboTwin2.0 dataset is also marked MIT. This cleaned derivative remains under MIT; attribution is retained in LICENSE.
- source manifest SHA256:
758dd41b41ce1836bfcc1ae2cd7858835f68d7e2e106bf3d0b7d4328d9678fff - clean manifest SHA256:
2f7a72c29cbce2c8b408b0a4c51977fc4976f374a115f9685c19f0e24aea7e3b - retained episodes: 26,617
- per-file hashes:
source/FILE_SHA256_SIZE.jsonl