Hi Guys,
Is there any embedding model from the huggingface where the size of the returned embedding with dimension 1536 as openai’s text-embedding-ada-002?
1 Like
There are many models that are too long or too short for that purpose.
Hey @John6666 . I have seen that blog. But the problem is that I dont have openai api keys. But I created my faiss_index with an dimension 1536. So from that blog i cause mistral’s model. But that model was very large where my streamlit application gets failed to download the model. Is there any other model? I have jina ai where the embedding dimension was 786.
I don’t know much about it, I’d have to do a search, but what about BAAI, dimension seems right and I don’t think it’s big.
Sorry, dimension is wrong…
1536 dimension models
Simply ordered by popularity
After all, it may be easiest to find from the MTEB leaderboard.
1 Like
Hey @John6666 . I have a doub that we cant use different embedding model in the time of vector store creation and in the inference. I mean i created faiss_index where i used openai embedding model where in the inference time i used other embedding models which matches the openai’s dimension size.
It’s not impossible to adjust to a lower dimension, apparently…
This topic was automatically closed 12 hours after the last reply. New replies are no longer allowed.