Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-19 22:23:50 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #3254] [BUG]: The similarity of the vector search results is abnormally high. #2093

New Issue
Closed
opened 2026-02-22 18:28:07 -05:00 by yindo · 2 comments
yindo commented 2026-02-22 18:28:07 -05:00
Owner
Copy Link
Copy Source

Originally created by @eiweorerw on GitHub (Feb 18, 2025).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3254

How are you running AnythingLLM?

Docker (local)

What happened?

The project I deployed using Docker has all configurations set to default, with the vector database being LanceDB by default. However, the data relevance returned by the /v1/workspace/fortest/vector-search interface does not match expectations. For instance, the vector data I uploaded pertains to the connections and differences between checks, promissory notes, and bills of exchange. When I asked, "What model and version are you using?", the vector results returned were 'distance': 0.7766722440719604, 'score': 0.9991058707237244. Similarly, when I inquired about the difference between checks and promissory notes, the vector results were 'distance': 0.6899669766426086, 'score': 0.996627688407898. This is perplexing because the relevance of the first query should be extremely low. What could be the reason for this discrepancy?

Are there known steps to reproduce?

deployed using Docker has all configurations set to default, with the vector database being LanceDB by default

Originally created by @eiweorerw on GitHub (Feb 18, 2025). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3254 ### How are you running AnythingLLM? Docker (local) ### What happened? The project I deployed using Docker has all configurations set to default, with the vector database being LanceDB by default. However, the data relevance returned by the /v1/workspace/fortest/vector-search interface does not match expectations. For instance, the vector data I uploaded pertains to the connections and differences between checks, promissory notes, and bills of exchange. When I asked, "What model and version are you using?", the vector results returned were 'distance': 0.7766722440719604, 'score': 0.9991058707237244. Similarly, when I inquired about the difference between checks and promissory notes, the vector results were 'distance': 0.6899669766426086, 'score': 0.996627688407898. This is perplexing because the relevance of the first query should be extremely low. What could be the reason for this discrepancy? ### Are there known steps to reproduce? deployed using Docker has all configurations set to default, with the vector database being LanceDB by default
yindo added the possible bug label 2026-02-22 18:28:07 -05:00
yindo closed this issue 2026-02-22 18:28:07 -05:00
yindo commented 2026-02-22 18:28:08 -05:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Feb 18, 2025):

Are you using "Accuract Optimized" in your workspace vector search preference? This would utilize reranking and would then show the reranking relevance.

Second to that, what embedder are you using? Is your document language and prompts in English or another language or multiple languages?

High accuracy in your semantic search would indicate reranking is enabled and you are getting the re-ranking relevance, not pure semantic search.

@timothycarambat commented on GitHub (Feb 18, 2025): Are you using "Accuract Optimized" in your workspace vector search preference? This would utilize reranking and would then show the reranking relevance. Second to that, what embedder are you using? Is your document language and prompts in English or another language or multiple languages? High accuracy in your semantic search would indicate reranking is enabled and you are getting the re-ranking relevance, not pure semantic search.
yindo commented 2026-02-22 18:28:08 -05:00
Author
Owner
Copy Link
Copy Source

@eiweorerw commented on GitHub (Feb 18, 2025):

Are you using "Accuract Optimized" in your workspace vector search preference? This would utilize reranking and would then show the reranking relevance.

Second to that, what embedder are you using? Is your document language and prompts in English or another language or multiple languages?

High accuracy in your semantic search would indicate reranking is enabled and you are getting the re-ranking relevance, not pure semantic search.

I have enabled the "Accuracy Optimized" option and am using the default embedder, AnythingLLM Embedder. The document language is Chinese. Is there a setting where I can specify the document language?

@eiweorerw commented on GitHub (Feb 18, 2025): > Are you using "Accuract Optimized" in your workspace vector search preference? This would utilize reranking and would then show the reranking relevance. > > Second to that, what embedder are you using? Is your document language and prompts in English or another language or multiple languages? > > High accuracy in your semantic search would indicate reranking is enabled and you are getting the re-ranking relevance, not pure semantic search. I have enabled the "Accuracy Optimized" option and am using the default embedder, AnythingLLM Embedder. The document language is Chinese. Is there a setting where I can specify the document language?
yindo changed title from [BUG]: The similarity of the vector search results is abnormally high. to [GH-ISSUE #3254] [BUG]: The similarity of the vector search results is abnormally high. 2026-06-05 14:44:30 -04:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
Desktop
Docker
Integration Request
Integration Request
OS: Linux
OS: Mobile
OS: Windows
UI/UX
blocked
bug
bug
core-team-only
documentation
duplicate
embed-widget
enhancement
feature request
github_actions
good first issue
investigating
needs info / can't replicate
possible bug
pull-request
Mirrored from GitHub Pull Request
 question
stage: specifications
wontfix
No Label possible bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#2093
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?