OpenGraphLabs-Research/ego-tactile-manipulation, measured from the public release on 13 September 2026. meta only, action column unreadable
OpenGraphLabs-Research/ego-tactile-manipulation carries 72 episodes and 463 distinct task strings, so there are more instructions than episodes, about 6.4 per episode. That is per segment annotation rather than per episode, which is richer than most releases and also means an episode count understates what is in there by roughly that factor.
| property | value |
|---|---|
| episodes | 72 |
| frames | 138,629 |
| frame rate | 30 fps |
| robot | ego_tactile_glove |
| action dimensions | |
| cameras | 1 |
| distinct task strings | 463 |
| shortest episode | 1,036 frames |
| median episode | 2,064 frames |
| longest episode | 4,369 frames |
The release is a meta directory plus one or more parquet files. Read data/chunk-000/file-000.parquet, group rows by the episode ranges in meta/episodes, and compute the same columns. No credentials.
This page is what a published release will tell you on its own. The full grading removes the part that scores worst and retrains on what is left, then 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.