chore(onnx): Remove dead code in find_scales and add test coverage

[shantoislamdev]

What does this PR do?

Type of change: Dead code cleanup, new tests

Overview: Remove confusing dead code in find_scales function and add test coverage.

The commented-out line # z = -np.round(temp).clip(...) in quant_utils.py represented a buggy implementation due to operator precedence. Python interprets -np.round(temp).clip(...) as -(np.round(temp).clip(...)), which incorrectly produces 0 for negative input ranges. The active code correctly performs negation before clipping.

This PR:

1. Removes the confusing commented-out dead code
2. Adds a parametrized test to verify the zero-point calculation works correctly for various input ranges

Testing

Added test_find_scales_zero_point which tests:

• Negative-only range
• Mixed range (negative to positive)
• Positive-only range
• Symmetric around zero

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?: Yes
• Did you add or update any necessary documentation?: No (code cleanup, no API changes)
• Did you update Changelog?: No (minor cleanup, not a feature/bug fix)

Additional Information

Files changed:

• modelopt/onnx/quantization/quant_utils.py: Removed 1 line of dead code
• tests/unit/onnx/test_quant_utils.py: Added test for find_scales

Summary by CodeRabbit

• Chores

  • Removed obsolete comment from quantization utilities code.

• Tests

  • Added test coverage for asymmetric quantization zero-point calculations to validate scale values, shape consistency, and zero-point bounds.

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

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[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

A comment line was removed from the find_scales function in the quantization utility module, and a new test case was added to validate asymmetric quantization zero-point calculations. These changes maintain existing functionality while improving test coverage.

Changes

Cohort / File(s)	Summary
Code Cleanup modelopt/onnx/quantization/quant_utils.py	Removed an unused/outdated comment from the zero-point branch within find_scales. No functional changes to computation logic.
Test Coverage tests/unit/onnx/test_quant_utils.py	Added new test function test_find_scales_zero_point to validate asymmetric quantization behavior, including positive scales, non-None zero-point, shape consistency, and zero-point bounds verification. Updated imports to include find_scales.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main changes: removing dead code (the commented line) from find_scales and adding test coverage for zero-point calculation.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

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

✨ Finishing touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

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.}