[GH-ISSUE #12231] feat: RAG: only list sources of cited sources #16517

Closed
opened 2026-04-19 22:25:22 -05:00 by GiteaMirror · 0 comments
Owner

Originally created by @almajo on GitHub (Mar 31, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/12231

Check Existing Issues

  • I have searched the existing issues and discussions.

Problem Description

When using a knowledge base with a model, the listed sources contain all seen document (chunks). However, even with k=5 we get 15 sources that have been given to the LLM. As most of these sources are not really relevant and the LLM only cites some of them, we should not confuse the enduser with a long list of sources, that are not related to the answer.

One additional problem I faced is that with 15 sources with 1000 tokens per chunk, loading the chat becomes slow. I'm not 100% sure it's related to this but it came up together.

Desired Solution you'd like

Only sources that are referenced by the llm should come up as sources. When writing a paper I would also only cite information I used, not everything I've read.

Alternatives Considered

No response

Additional Context

No response

Originally created by @almajo on GitHub (Mar 31, 2025). Original GitHub issue: https://github.com/open-webui/open-webui/issues/12231 ### Check Existing Issues - [x] I have searched the existing issues and discussions. ### Problem Description When using a knowledge base with a model, the listed sources contain all seen document (chunks). However, even with k=5 we get 15 sources that have been given to the LLM. As most of these sources are not really relevant and the LLM only cites some of them, we should not confuse the enduser with a long list of sources, that are not related to the answer. One additional problem I faced is that with 15 sources with 1000 tokens per chunk, loading the chat becomes slow. I'm not 100% sure it's related to this but it came up together. ### Desired Solution you'd like Only sources that are referenced by the llm should come up as sources. When writing a paper I would also only cite information I used, not everything I've read. ### Alternatives Considered _No response_ ### Additional Context _No response_
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/open-webui#16517