Fix log analysis agent: Addressed Parsing crash and model incompatibility issues#411
Open
venkatesh71097 wants to merge 3 commits intoNVIDIA:mainfrom
Open
Fix log analysis agent: Addressed Parsing crash and model incompatibility issues#411venkatesh71097 wants to merge 3 commits intoNVIDIA:mainfrom
venkatesh71097 wants to merge 3 commits intoNVIDIA:mainfrom
Conversation
…nd AttributeError
venkatesh71097
changed the title
venkatesh71097
changed the title
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
🛠️ Fix Log Analysis Agent: Robust Parsing & Bug Fixes
📋 Summary
This PR addresses critical stability issues in the log analysis agent caused by model incompatibility and output parsing failures. It specifically fixes the
AttributeError: 'NoneType'crash, handles reasoning models (likellama-3.3-nemotron) gracefully, and resolves Pydantic deprecation warnings.🐛 Issues Fixed
AttributeError: 'NoneType' object has no attribute 'binary_score'occurring when the model output parser failed.<think>blocks (reasoning traces) broke the standard JSON parser.- Deprecation Warnings: Resolved
LangChainDeprecationWarningrelated topydantic_v1.🚀 Changes
| Component | Change Description |
|-----------|--------------------|
| utils.py | Implemented a robust regex-based parsing pipeline. It now aggressively strips
<think>reasoning tags before parsing JSON, ensuring stability even with "chatty" or reasoning-heavy models. Restored missingimport json. || graphnodes.py | Added defensive coding with
.get()access and null checks to handle graceful degradation if a grader returnsNone. || graphedges.py | Updated logic to correctly handle dictionary outputs instead of expecting Pydantic objects. |
| binary_score_models.py | Migrated to direct
pydantic(v2) imports. || .gitignore | Added ignore rules for .env, debug_log.txt, and
.venv/. |