add imatrix calibration data resolution to quant config

[Roderick-Wu]

vllm-project/llm-compressor#2920

[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.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 7c77315e-5adf-4a65-a797-3dbef6cdbd9f

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

Use the checkbox below for a quick retry:

• 🔍 Trigger review

Walkthrough

Changes

Calibration Detection

Layer / File(s)	Summary
Weight observer calibration check src/compressed_tensors/quantization/quant_config.py	requires_calibration_data now returns True when a scheme has a weight observer whose lowercase value is imatrix_mse.

Estimated code review effort: 1 (Trivial) | ~5 minutes

Suggested reviewers: kylesayrs, brian-dellabetta

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title matches the main change: adding imatrix-related calibration handling in quant config.
Description check	✅ Passed	The description is only a link, but it points to the related upstream PR for this change.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch Roderick-Wu/imatrix-fix

---

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.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@src/compressed_tensors/quantization/quant_config.py`:
- Around line 300-305: Replace the boolean-wrapped any() condition in the
surrounding scheme loop with a direct if condition checking scheme.weights,
scheme.weights.observer, and its case-insensitive "imatrix_mse" value; preserve
the existing return True behavior when all checks match.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 582515c3-a4f2-4fe9-8063-19b202b49815

📥 Commits

Reviewing files that changed from the base of the PR and between bf81005 and 5875776.

📒 Files selected for processing (1)

• src/compressed_tensors/quantization/quant_config.py

🔗 Linked repositories identified

CodeRabbit considers these linked repositories for cross-repo context during reviews:

• vllm-project/llm-compressor (manual) → reviewed against open PR #2920 Roderick-Wu/remove-imatrixgatherer instead of the default branch

[coderabbitai]

🎯 Functional Correctness | 🔴 Critical | ⚡ Quick win

Replace any() with a direct condition

any() is receiving a boolean, not an iterable, so every call reaching this branch raises TypeError: 'bool' object is not iterable. The surrounding loop already evaluates one scheme at a time; use a direct if condition instead.

Proposed fix

-            if any(
+            if (
                 scheme.weights is not None
                 and scheme.weights.observer is not None
                 and scheme.weights.observer.lower() == "imatrix_mse"
             ):

📝 Committable suggestion

‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.

Suggested change

	if any(
	scheme.weights is not None
	and scheme.weights.observer is not None
	and scheme.weights.observer.lower() == "imatrix_mse"
	):
	return True
	if (
	scheme.weights is not None
	and scheme.weights.observer is not None
	and scheme.weights.observer.lower() == "imatrix_mse"
	):
	return True

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/compressed_tensors/quantization/quant_config.py` around lines 300 - 305,
Replace the boolean-wrapped any() condition in the surrounding scheme loop with
a direct if condition checking scheme.weights, scheme.weights.observer, and its
case-insensitive "imatrix_mse" value; preserve the existing return True behavior
when all checks match.

Source: Path instructions

[HDCharles]

lets make a list of observers that require calibration in llm-compressor and use it here instead of hardcoding a bunch of observer names

[mergify]

Merge Protections

🔴 1 of 1 protections blocking · waiting on 👀 reviews

	Protection	Waiting on
🔴	Require one maintainer review	👀 reviews

🔴 Require one maintainer review

Waiting for

• #changes-requested-reviews-by = 0
• any of:
  • approved-reviews-by=HDCharles
  • approved-reviews-by=brian-dellabetta
  • approved-reviews-by=dsikka
  • approved-reviews-by=kylesayrs

This rule is failing.

All PRs must have at least one approving review from a maintainer before merging.

• #changes-requested-reviews-by = 0
• any of:
  • approved-reviews-by=HDCharles
  • approved-reviews-by=brian-dellabetta
  • approved-reviews-by=dsikka
  • approved-reviews-by=kylesayrs