Module redlite.dataset.json_dataset

Classes

JSONDataset

class JSONDataset(
    *,
    path: str,
    name: str,
    split: Literal['test', 'train'],
    labels: dict[str, str] | None = None
)

Dataset from a local JSONL file.

Each line of the dataset file must be a DatasetItem serialized to JSON representation. File must use UTF-8 encoding. There should be no BOM-markers (some Microsoft tools produce those).

• path (str): Location of JSONL file.
• name (str): Dataset name.
• split (str): Dataset split.
• labels (dict[str, str]): Labels.

Ancestors (in MRO)

• redlite.dataset.memory_dataset.MemoryDataset
• redlite.dataset._load.ValidatingDataset
• redlite._core.NamedDataset
• collections.abc.Sized
• collections.abc.Iterable