Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-19 14:13:35 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #2525] [Question] Why is the search result quality of anythingllm in development mode much less than that of AnythingLLM Desktop? #1633

New Issue
Closed
opened 2026-02-22 18:25:48 -05:00 by yindo · 1 comment
yindo commented 2026-02-22 18:25:48 -05:00
Owner
Copy Link
Copy Source

Originally created by @fengnex on GitHub (Oct 23, 2024).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/2525

Ollama, the same local LLM llama3.1:latest, lancedb and nomic-embed-text:latest are used under the two environments as described in the above title, then save the same pdf and docx documents and search the same question against both anythingllms, but the quality of the answers given by the two is completely different：
The answers provided by the desktop version are generally comprehensive in content, formal in format, and high-end. However, the same questions are not even answered in development mode, or the answers are very brief, and often lack some important content, usually without much format!

What are the reasons for this huge difference? How to ensure that the version compiled by oneself in development mode can also achieve the same or similar query quality as the desktop version? Can anyone please explain these?

Originally created by @fengnex on GitHub (Oct 23, 2024). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/2525 Ollama, the same local LLM `llama3.1:latest`, lancedb and `nomic-embed-text:latest` are used under the two environments as described in the above title, then save the same pdf and docx documents and search the same question against both anythingllms, but the quality of the answers given by the two is completely different： The answers provided by the desktop version are generally comprehensive in content, formal in format, and high-end. However, the same questions are not even answered in development mode, or the answers are very brief, and often lack some important content, usually without much format! What are the reasons for this huge difference? How to ensure that the version compiled by oneself in development mode can also achieve the same or similar query quality as the desktop version? Can anyone please explain these?
yindo closed this issue 2026-02-22 18:25:48 -05:00
yindo commented 2026-02-22 18:25:48 -05:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Oct 24, 2024):

My intuitive reaction is that this is ultimately the difference from the inherent difference between runs of the same models. However, there is one major detail missing that matters.

Are you running the desktop local LLM using an external Ollama and using that same external Ollama for development mode as well? There is very well a way to get the exact message array passed to the Ollama instance. Otherwise, this could be differences between any number of things. I can assure you the code, logic, and ultimately the messages passed to the model should be the exact same which puts the root cause outside of the software and onto the inferencing process.

Interesting question, looking forward to seeing what is going on here - closing but please continue the conversation with more information.

@timothycarambat commented on GitHub (Oct 24, 2024): My intuitive reaction is that this is ultimately the difference from the inherent difference between runs of the same models. However, there is one major detail missing that matters. Are you running the desktop local LLM using an _external_ Ollama and using that _same_ external Ollama for development mode as well? There is very well a way to get the exact message array passed to the Ollama instance. Otherwise, this could be differences between any number of things. I can assure you the code, logic, and ultimately the messages passed to the model should be _the exact same_ which puts the root cause outside of the software and onto the inferencing process. Interesting question, looking forward to seeing what is going on here - closing but please continue the conversation with more information.
yindo changed title from [Question] Why is the search result quality of anythingllm in development mode much less than that of AnythingLLM Desktop? to [GH-ISSUE #2525] [Question] Why is the search result quality of anythingllm in development mode much less than that of AnythingLLM Desktop? 2026-06-05 14:41:50 -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
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#1633
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?