src.utils.dataset_factory module

src.utils.dataset_factory.make_dataset_without_config(repo_id: str, action_delta_indices: List, observation_delta_indices: List | None = None, root: str | None = None, video_backend: str = 'pyav', episodes: list[int] | None = None, revision: str | None = None, use_imagenet_stats: bool = True) → LeRobotDataset | MultiLeRobotDataset

Handles the logic of setting up delta timestamps and image transforms before creating a dataset.

Args:

cfg (TrainPipelineConfig): A TrainPipelineConfig config which contains a DatasetConfig and a PreTrainedConfig.

Raises:

NotImplementedError: The MultiLeRobotDataset is currently deactivated.

Returns:

LeRobotDataset | MultiLeRobotDataset

src.utils.dataset_factory.resolve_delta_timestamps_without_config(ds_meta: LeRobotDatasetMetadata, action_delta_indices: List, observation_delta_indices: List | None = None) → dict[str, list] | None

Resolves delta_timestamps by reading from the ‘delta_indices’ properties of the PreTrainedConfig.

Args:

cfg (PreTrainedConfig): The PreTrainedConfig to read delta_indices from. ds_meta (LeRobotDatasetMetadata): The dataset from which features and fps are used to build

delta_timestamps against.

Returns:

dict[str, list] | None: A dictionary of delta_timestamps, e.g.:

{

“observation.state”: [-0.04, -0.02, 0] “observation.action”: [-0.02, 0, 0.02]

} returns None if the the resulting dict is empty.