Updated 2026-10-15 · Rechecked with every build

Robot Training Data Statistics

Numbers from Datum's checks over the 1000 most downloaded LeRobot datasets on the Hugging Face Hub. Every stat stands alone. Cite freely with a link.

Stat 1Of the 1000 most downloaded LeRobot datasets, 54% fail at least one basic quality check.

Stat 2221 of 1000 datasets (22%) are junk: missing metadata, fewer than 5 episodes, or placeholder task labels.

Stat 3281 of 1000 datasets (28%) have no usable license, so nobody can safely build on them.

Stat 4199 of 1000 datasets have no readable meta/info.json file.

Stat 5463 of 1000 datasets (46%) pass every check Datum can run from metadata.

Stat 6The most common robot in the checked set is Franka, with 271 datasets that pass the junk gate.

Stat 7641 of 1000 checked datasets use the new LeRobot v3.0 format.

Stat 8The checked datasets hold 2,032,243 episodes in total.

Stat 9120 datasets declare no camera streams at all in their schema.

Stat 10693 of 1000 datasets (69%) carry a license that allows commercial use.