Conversation
…ment - Added `_attach_model_profile` function to attach model context metadata to `ChatLiteLLM`. - Updated `create_chat_litellm_from_config` and `create_chat_litellm_from_agent_config` to utilize the new profile attachment. - Improved context profile caching in `llm_router_service.py` to include both minimum and maximum input tokens, along with token model names for better context management. - Introduced new methods for token counting and context trimming based on model profiles.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
There was a problem hiding this comment.
Review by RecurseML
🔍 Review performed on 5571e8a..eec4db4
✨ No bugs found, your code is sparkling clean
✅ Files analyzed, no issues (1)
• surfsense_backend/app/services/llm_router_service.py
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
…ment
_attach_model_profilefunction to attach model context metadata toChatLiteLLM.create_chat_litellm_from_configandcreate_chat_litellm_from_agent_configto utilize the new profile attachment.llm_router_service.pyto include both minimum and maximum input tokens, along with token model names for better context management.Description
Motivation and Context
FIX #
Screenshots
API Changes
Change Type
Testing Performed
Checklist
High-level PR Summary
This PR enhances LLM configuration and routing by introducing model profile attachment and context-aware message trimming capabilities. The changes add a
_attach_model_profilefunction that captures model context metadata (likemax_input_tokens) from LiteLLM's model info and attaches it toChatLiteLLMinstances. The router service is updated to cache both minimum and maximum input token limits across all deployments, enabling smarter context management. Most significantly, this introduces sophisticated context trimming logic that uses binary search to intelligently truncate large messages (especially tool responses and document context) when they exceed the model's context window, preferring XML document boundaries for cleaner cuts and preserving system messages to maintain agent instructions.⏱️ Estimated Review Time: 30-90 minutes
💡 Review Order Suggestion
surfsense_backend/app/agents/new_chat/llm_config.pysurfsense_backend/app/services/llm_router_service.py