vertebrae.extractors.multi_output

Generic multi-output extractor wrapper.

Classes

MultiOutputExtractor

Wrap a callable that returns multiple named embedding matrices.

Module Contents

class vertebrae.extractors.multi_output.MultiOutputExtractor(name, output_specs, transform_many_fn, fit_fn=None, modality='unknown', extractor_type='custom_multi_output', recipe_data=None, allow_sparse=True, streaming_safe=False, cache_identity=None)[source]

Wrap a callable that returns multiple named embedding matrices.

Parameters:

  • name (str) – User-facing parent extractor name.

  • output_specs (List[vertebrae.extractors.base.EmbeddingOutputSpec]) – Named output descriptions.

  • transform_many_fn (Callable[[Any], Any]) – Callable returning a dict or sequence of embedding outputs.

  • fit_fn (Optional[Callable[[Any, Any], Any]]) – Optional callable invoked during fit.

  • modality (str) – Input modality metadata.

  • extractor_type (str) – Extractor family metadata.

  • recipe_data (Optional[Dict[str, Any]]) – Extra serializable data captured in recipe().

  • allow_sparse (bool) – Whether sparse outputs are accepted.

  • streaming_safe (bool) – Whether independent batches can be transformed safely.

  • cache_identity (Optional[str])