lance-format/pusht-lance

pusht, stored in Lance

lerobot/pusht converted to the Lance layout that LeRobot's LanceDBDataset reads natively. Same meta/ sidecar as the original, tabular features in frames.lance, encoded videos in a videos.lance blob v2 column.

from lerobot.datasets import LeRobotDataset  # not needed, shown for contrast
from lerobot.datasets import LanceDBDataset

ds = LanceDBDataset("lance-format/pusht-lance")  # streams tables from the Hub, only meta/ is downloaded
item = ds[0]

Or just point training at it. The dataset factory detects the format automatically:

lerobot-train --dataset.repo_id=lance-format/pusht-lance ...

Requires lancedb>=0.35.0b2 (pip install --pre --extra-index-url https://pypi.fury.io/lancedb/ lancedb) and the lance-dataloader branch of LeRobot until the upstream PR lands.