Leodocr/egovista-kitchen-manipulation-v1, measured from the public release on
16 September 2026.
Every row is read out of the release. Nothing here is estimated and nothing is rounded up.
| episodes | 4 |
| frames | 2,662 |
| action columns | none |
| what the release carries instead | observation.pose, observation.hand_left, observation.hand_right, segmentation masks, contact flags |
| distinct task strings | 4 |
| what those task strings are | the integers 0, 1, 2 and 3 |
The release is laid out as LeRobot and a loader will open it, then there is nothing to regress onto. It is an observation corpus, which is a fine thing to be, but a buyer who assumes LeRobot layout means LeRobot semantics finds out after they have paid.
Read data/chunk-000/file-000.parquet and list the column names. No credentials.
This page is what the release tells you on its own. A full grading removes the part that scores worst, retrains on what is left, and reports whether the smaller corpus scores higher, with rollout counts and intervals. The one result measured on physical hardware is Stera-10M, two arms on a four arm SO-ARM101 bench, 21.9% against 36.9% task success across 1,270 rollouts. The Stera-10M report.