jinaai

/

jina-embeddings-v4

like 414

Follow

Jina AI 1.44k

Hi @Kong-Mei , do you want to quantize the whole model or just the embeddings? We will soon publish quantized versions for INT8 and Binary embeddings.

Thanks! I only need the embeddings for a multimodal retrieval task. By the way, may I ask how much speed improvement we can expect from quantization?
If possible, I’d like to quantize both qwen2.5-VL-3b and the multi_vector_projector.

here u go: https://github.com/jina-ai/jina-embeddings-v4-gguf

Thanks!