[AutoRound] Update autoround to the release version

[yiliu30]

SUMMARY:

• Pin autoround to the 0.9.1
• Minor fix for quantize_block
• Expose the batch_size to users
  TEST PLAN:

pytest -svv ./tests/llmcompressor/transformers/autoround/test_autoround_oneshot.py 

cc @hshen14 @thuang6 @chensuyue

[gemini-code-assist]

Summary of Changes

Hello @yiliu30, 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 primarily focuses on stabilizing the autoround integration by pinning its dependency to the official 0.9.0 release. Additionally, it includes a small but important fix to ensure proper device parameter handling within the quantize_block function, which improves robustness and compatibility with the updated autoround library.

Highlights

• Dependency Update: The auto-round dependency has been updated from a specific git branch reference to its official release version 0.9.0 in setup.py.
• Bug Fix: A minor fix was applied in src/llmcompressor/modifiers/autoround/base.py to explicitly convert the device parameter to a string before passing it to the quantize_block function, ensuring correct type handling.

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 updates the autoround dependency to its 0.9.0 release version, replacing the previous git-based dependency. This is a good practice for ensuring stability and reproducibility. The corresponding change in src/llmcompressor/modifiers/autoround/base.py, which converts the torch.device object to a string when calling quantize_block, is a necessary API adaptation for the new library version. The changes are straightforward and correct.

[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.

[kylesayrs]

Why is this batch size argument changed? Does this argument actually do anything, given that calibration is done by LC pipelines, not AutoRound?

[yiliu30]

Yes, it does. The calibration samples collected by LC are first stacked into a list, and then AutoRound resamples data from that list based on the batch_size here.