search

huggingface / text-embeddings-inference

A blazing fast inference solution for text embeddings models
https://huggingface.co/docs/text-embeddings-inference/quick_tour
Apache License 2.0
2.5k stars 157 forks source link

Support Colbert v2 #127

Open dongxiaolong opened 7 months ago

dongxiaolong commented 7 months ago

Model description

ColBERT is a fast and accurate retrieval model, enabling scalable BERT-based search over large text collections in tens of milliseconds. ColBERT github ColBERT huggingface

Open source status

Provide useful links for the implementation

No response

OlivierDehaene commented 7 months ago

What do you need exactly for it to be supported? Is supporting the embeddings per token with compression enough?

dongxiaolong commented 7 months ago

What do you need exactly for it to be supported? Is supporting the embeddings per token with compression enough?

I appreciate your response. I believe it would be better to support both the colbert embedding model and the reranker model. Here are two distinct applications. In llamaindex and langchain as a embedding model: end2end retrieval ragatouille In this blog as a reranker model:colbert as a re-ranker

hiepxanh commented 6 months ago

colbert is very intersting, I want to try it to on HF text embedding inference