Strange values for title conversations (LLM issue) #7448
-
What happened?It's been fun getting LibreChat set up so far, but I've noticed some strange behavior when it tries to generate conversation titles. It seems as though they are all fragments of math questions:
Version Informationghcr.io/danny-avila/librechat-dev latest 1140153e2f80 2 days ago 1.03GB Steps to Reproduce
What browsers are you seeing the problem on?No response Relevant log outputsee screenshot belowScreenshots
Code of Conduct
|
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 1 reply
-
|
I've tried with three models from LM Studio and noticed the same issue:
|
Beta Was this translation helpful? Give feedback.
-
|
It's an LLM issue, not an application issue. Once you can customize the title strategy, you could mitigate this by tailoring specific prompts for specific models, as right now the current strategy doesn't always work well with smaller models. |
Beta Was this translation helpful? Give feedback.
-
|
Makes sense, thanks! |
Beta Was this translation helpful? Give feedback.
It's an LLM issue, not an application issue. Once you can customize the title strategy, you could mitigate this by tailoring specific prompts for specific models, as right now the current strategy doesn't always work well with smaller models.