AWQ per-channel fix

[zhanglei1172]

SUMMARY:
"Fix AWQ modifier to support per-channel quantization configuration"

TEST PLAN:
"Previously, AWQ modifier defaulted to using group quantization's group_size configuration, which caused errors when per-channel quantization configuration was passed in."

[github-actions]

👋 Hi! Thank you for contributing to llm-compressor. Please add the ready label when the PR is ready for review.

Note: This is required to complete the testing suite, please only add the label once the PR is code complete and local testing has been performed.

[gemini-code-assist]

Summary of Changes

Hello @zhanglei1172, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request addresses a critical issue in the AWQ modifier where it previously failed to support per-channel quantization configurations. The changes ensure that the system properly recognizes and adapts its behavior when per-channel quantization is intended, specifically by adjusting how _group_size is interpreted and by conditionally applying group-specific weight smoothing operations. This fix allows for more flexible and robust quantization strategies within the framework.

Highlights

• Per-channel Quantization Support: The AWQ modifier now correctly handles per-channel quantization configurations by explicitly setting _group_size to -1 when no group size is specified, indicating a per-channel setup.
• Conditional Weight Smoothing: The weight smoothing logic within the _apply_smoothing method has been updated to only execute when a positive _group_size is defined. This prevents errors when per-channel quantization is active, as group-based operations are not applicable in such cases.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request addresses an issue in the AWQ modifier where the group size was not correctly handled when per-channel quantization was used. The changes ensure that the group size is set to -1 when it's None, and a conditional statement is added to apply smoothing only when the group size is greater than 0. This resolves errors that occurred when per-channel quantization configurations were passed in. The review focuses on the correctness and maintainability of the changes.

[brian-dellabetta]

Thanks for the contribution, but I don't think this will work correctly as is. See change request.

Have you been able to run AWQ with channel-wise quant on this branch and seen improvement over simple round-to-nearest channel-wise quant?