Skip to content

lit_deltalake.datasets.conversion

source module lit_deltalake.datasets.conversion

Functions

  • dataclass_columns — Return Delta column names declared as dataclass fields.

  • apply_transforms — Apply one or more ordered whole-sample transforms.

  • to_sample — Convert one selected record into a mapping or typed dataclass sample.

  • dataclass_collate — Collate matching dataclass samples while retaining their type.

  • tensordict_collate — Collate mappings or dataclasses into one TensorDict batch.

source dataclass_columns(row_type: RowType) → tuple[str, ...]

Return Delta column names declared as dataclass fields.

Raises

  • InvalidDatasetConfigurationError

source apply_transforms(sample: object, transform: Transform | None) → object

Apply one or more ordered whole-sample transforms.

source to_sample(record: Record, row_type: RowType | None, transform: Transform | None) → SampleT

Convert one selected record into a mapping or typed dataclass sample.

source dataclass_collate(samples: list[object]) → object

Collate matching dataclass samples while retaining their type.

source tensordict_collate(samples: list[object]) → object

Collate mappings or dataclasses into one TensorDict batch.

Raises

  • InvalidDatasetConfigurationError

  • BackendUnavailableError