[ONNX][Autocast] Minor bug fixes (AI-assisted)

[galagam]

What does this PR do?

Type of change: Bug fix

Overview:
Conducted a thorough code review for AutoCast using ClaudeCode to identify bugs and improve code quality.
Fixed some low hanging issues.

Implemented fixes:

• DepthReductionRule: normalize negative axis and check for valid axis range
• Init scale_dimension in GraphSanitizer to None (was left uninitialized)
• Simplify logic around custom_ops_low_precision_nodes and prevent duplicate nodes

Before your PR is "Ready for review"

• Make sure you read and follow Contributor guidelines and your commits are signed.
• Is this change backward compatible?: Yes
• Did you write any new necessary tests?: No
• Did you add or update any necessary documentation?: No
• Did you update Changelog?: No

Additional Information

Summary by CodeRabbit

• Bug Fixes

  • Fixed potential undefined variable error in layer normalization pattern detection.

• Improvements

  • Optimized low-precision node classification to eliminate duplicate processing, improving efficiency of the autocast system.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

Important

Review skipped

Auto incremental reviews are disabled on this repository.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

• 🔍 Trigger a full review

📝 Walkthrough

Walkthrough

Two internal utility modules are updated: a variable initialization fix in the graph sanitizer to prevent undefined-variable usage, and a deduplication optimization in the node classifier that uses a set-based approach to prevent reprocessing of low-precision nodes.

Changes

Cohort / File(s)	Summary
Graph Sanitizer modelopt/onnx/autocast/graphsanitizer.py	Initialize scale_dimension = None before final_node assignment to guarantee the variable is always defined before use
Node Classifier modelopt/onnx/autocast/nodeclassifier.py	Deduplicate low-precision nodes during classification by using a set to track processed nodes; skip reprocessing nodes already in the low-precision set; convert set back to list before returning to maintain public API signature

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~12 minutes

🚥 Pre-merge checks | ✅ 2 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 66.67% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title '[ONNX][Autocast] Minor bug fixes (AI-assisted)' is directly related to the changeset, which contains bug fixes in ONNX autocast modules (graphsanitizer.py and nodeclassifier.py). It accurately summarizes the main change.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[codecov]

Codecov Report

❌ Patch coverage is 54.54545% with 5 lines in your changes missing coverage. Please review.
✅ Project coverage is 74.01%. Comparing base (2c73de0) to head (7958d38).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
modelopt/onnx/autocast/nodeclassifier.py	50.00%	5 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #822      +/-   ##
==========================================
- Coverage   74.02%   74.01%   -0.02%     
==========================================
  Files         192      192              
  Lines       19664    19672       +8     
==========================================
+ Hits        14557    14560       +3     
- Misses       5107     5112       +5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.