[NVFp4][MoE Calibration] Fix MoE Calibration Context

[dsikka]

SUMMARY:

• Fixes bug where despite being calibrated within the MoE Calibration context, some experts were not being calibrated / activation scales remained changed. The flag should be set to True by default
• Update the Qwen3 MoE NVFP4 Example to only use 20 calibration samples, not 200

[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 @dsikka, 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 bug in the Mixture of Experts (MoE) calibration context, ensuring that all experts are properly calibrated by default. This change enhances the reliability of MoE model quantization. Additionally, it optimizes the Qwen3 MoE NVFP4 example by significantly reducing the number of calibration samples, leading to faster execution of the example.

Highlights

• MoE Calibration Bug Fix: Corrected an issue in the moe_calibration_context where not all experts were being calibrated, even when within the calibration context. This is resolved by setting the calibrate_all_experts parameter to True by default, ensuring comprehensive calibration.
• Qwen3 MoE NVFP4 Example Optimization: The Qwen3 MoE NVFP4 example has been updated to use a reduced number of calibration samples, specifically 20 samples instead of 200, to streamline the calibration process and improve efficiency.

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 a bug in Mixture-of-Experts (MoE) calibration, ensuring all experts are correctly calibrated by default. The core change adjusts the calibrate_all_experts parameter's default value to True within the moe_calibration_context, which is a sensible and correct fix. The accompanying reduction in calibration samples for the Qwen3 example is a practical adjustment to maintain example performance. My review includes suggestions to add a docstring for the modified function and to clarify the purpose of the low number of calibration samples in the example, both aimed at improving code clarity and maintainability.