[Bugfix] infer_quantization_format when model only has activation quantization

[kylesayrs]

Purpose

• Fix KV cache tests, whose models only have activation quantization

Background

Previously, is_model_quantized would only check for quantization on leaf modules. Now it checks on attention modules as well, but since we have examples of attention modules with only activation quantization, this triggers a bug in infer_quantization_format

Testing

• Requires [Bugfix] Only quant-compress modules with weight quantization neuralmagic/compressed-tensors#387 to pass KV cache tests

[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 @kylesayrs, 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 specific bug in the infer_quantization_format utility function within the llmcompressor library. The primary goal is to ensure that the function correctly identifies and infers the quantization format for models that exclusively utilize activation quantization, which was previously causing failures, particularly in KV cache tests. The changes simplify the initial checks for quantization presence within the function.

Highlights

• Bugfix for infer_quantization_format: The infer_quantization_format function was updated to correctly handle models that only have activation quantization. This resolves a bug that was exposed by recent changes to is_model_quantized, which now correctly identifies quantization in attention modules, leading to issues with KV cache tests.
• Refactor infer_quantization_format logic: The internal logic of infer_quantization_format was streamlined. It now first determines unique quantization arguments for weights and inputs, and then explicitly returns None if no weights are found to be quantized. This change replaces a previous check using is_model_quantized and ensures more robust behavior.

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 in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

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 issue 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 is currently in preview and 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 to provide feedback.

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

The pull request fixes a bug in infer_quantization_format for models with only activation quantization. A suggestion has been provided to improve efficiency by avoiding an expensive function call in certain cases.