MultimodalEmbeddingModel

A loaded dual-tower embedding model: texts and images share one space.

embedding_dim (property)

tokenizer (property)

embed

embed(self, texts: 'str | Sequence[str]') -> 'clika_runtime.Tensor'

embed_image

embed_image(self, image: 'clika_runtime.Tensor') -> 'clika_runtime.Tensor'

A Float32 [D] tensor for one [H, W, 3] image.