DynamicIntelligence/humanoid-robots-training-dataset, measured from the public release on 13 September 2026. frame level, every row read
We read every frame of DynamicIntelligence/humanoid-robots-training-dataset, 149 episodes and 60,485 rows, and 2 of its episode files have no action column in them at all, data/chunk-001/episode_000127.parquet, data/chunk-001/episode_000141.parquet. The same read also found that meta declares 61017 frames, the data has 60485 rows. A training run that globs the data directory will either fail on those files or silently train on fewer episodes than the release claims.
| property | value |
|---|---|
| episodes | 149 |
| frames | 60,485 |
| frame rate | 30 fps |
| robot | unknown |
| action dimensions | 24 |
| cameras | |
| distinct task strings | |
| shortest episode | not declared |
| median episode | not declared |
| longest episode | not declared |
A trailing hold is the number of frames at the end during which the commanded action does not change by more than 1e-4 on any component. Seconds come from the frame rate the release declares.
| trailing hold at least | episodes | share |
|---|---|---|
| 1 frames , 0.03 seconds | 0 | 0.0% |
| 5 frames , 0.17 seconds | 0 | 0.0% |
| 10 frames , 0.33 seconds | 0 | 0.0% |
| 30 frames , 1.0 seconds | 0 | 0.0% |
| 60 frames , 2.0 seconds | 0 | 0.0% |
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.