[GH-ISSUE #4327] [BUG]: bge-m3 embedding model got 0 stored into vector store #2753

Closed
opened 2026-02-22 18:31:06 -05:00 by yindo · 1 comment
Owner

Originally created by @vincentsunx on GitHub (Aug 25, 2025).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/4327

How are you running AnythingLLM?

Docker (local)

What happened?

I have used bge-m3 as my embedding model.
My embedding model can worke correctly, it can generate vectors successfully.

Image

I got no errors and no warnings and inert into vector store successfully. However, vectors stored in my vector database are all 0 which not works for RAG.

Image

Is that the problem for parsing requsets, cause I tried other embedding models, it works correctly.

Are there known steps to reproduce?

No response

Originally created by @vincentsunx on GitHub (Aug 25, 2025). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/4327 ### How are you running AnythingLLM? Docker (local) ### What happened? I have used bge-m3 as my embedding model. My embedding model can worke correctly, it can generate vectors successfully. <img width="1135" height="261" alt="Image" src="https://github.com/user-attachments/assets/e646c148-e72e-465e-877a-01397a21af89" /> I got no errors and no warnings and inert into vector store successfully. However, vectors stored in my vector database are all 0 which not works for RAG. <img width="1168" height="446" alt="Image" src="https://github.com/user-attachments/assets/cff3ef76-c109-443f-88c7-159dd07b35ca" /> Is that the problem for parsing requsets, cause I tried other embedding models, it works correctly. ### Are there known steps to reproduce? _No response_
yindo added the possible bug label 2026-02-22 18:31:06 -05:00
yindo closed this issue 2026-02-22 18:31:06 -05:00
Author
Owner

@timothycarambat commented on GitHub (Sep 12, 2025):

How are you viewing data stored in the vector store? Most vector stores dont return embeddings unless you explicitly ask for embeddings: true in your request because they are often very long and verbose and increase the size of HTTP responses dramatically.

For example, in Chroma - https://docs.trychroma.com/docs/querying-collections/query-and-get#choosing-which-data-is-returned you need to specify it. The above are not screenshots from AnythingLLM or our API - so im not sure how this is a bug at all

@timothycarambat commented on GitHub (Sep 12, 2025): How are you viewing data stored in the vector store? Most vector stores dont return embeddings unless you explicitly ask for `embeddings: true` in your request because they are often very long and verbose and increase the size of HTTP responses dramatically. For example, in Chroma - https://docs.trychroma.com/docs/querying-collections/query-and-get#choosing-which-data-is-returned you need to specify it. The above are not screenshots from AnythingLLM or our API - so im not sure how this is a bug at all
yindo changed title from [BUG]: bge-m3 embedding model got 0 stored into vector store to [GH-ISSUE #4327] [BUG]: bge-m3 embedding model got 0 stored into vector store 2026-06-05 14:48:21 -04:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#2753