Request for model jinaai/jina-embeddings-v2-base-de

Hi there,

Since the jinaai/jina-embeddings-v2-base-de model is one of the most capable models for german text, I'd love to see it supported.

Is there anything I can help you with?

Hey @deichrenner

I think it should be as easy as adding it to

fastembed/fastembed/text/jina_onnx_embedding.py

Line 10 in 615d6ee

supported_jina_models = [

Plus,

fastembed/tests/test_text_onnx_embeddings.py

Line 8 in 615d6ee

CANONICAL_VECTOR_VALUES = {

To compute a canonical vector, we usually take the same model from HF transformers, apply it to some test input, take 5 first values and round to 4 digits.

Would you like to make a contribution?

Hi @joein,
Yes, the code changes are reasonably simple. My biggest question was about the quantized model. Who manages those?
Sure, I'd be more than happy to contribute.