Return the embedding dimension for a given modality. Subclasses should override this to return the actual dimension so that auto-generated FeatureView schemas use the correct vector_length. Args: modality: The modality to query (e.g. "text", "image").
(self, modality: str)
| 54 | return list(self._modality_handlers.keys()) |
| 55 | |
| 56 | def get_embedding_dim(self, modality: str) -> Optional[int]: |
| 57 | """ |
| 58 | Return the embedding dimension for a given modality. |
| 59 | |
| 60 | Subclasses should override this to return the actual dimension |
| 61 | so that auto-generated FeatureView schemas use the correct vector_length. |
| 62 | |
| 63 | Args: |
| 64 | modality: The modality to query (e.g. "text", "image"). |
| 65 | |
| 66 | Returns: |
| 67 | The embedding dimension, or None if unknown. |
| 68 | """ |
| 69 | return None |
| 70 | |
| 71 | @abstractmethod |
| 72 | def embed(self, inputs: List[Any], modality: str) -> "np.ndarray": |
no outgoing calls
no test coverage detected