[NVFP4] Add lm-eval test case

[dsikka]

Summary

• Enable and nvfp4 weekly lm-eval test

vLLM:

|Tasks|Version|     Filter     |n-shot|  Metric   |   |Value |   |Stderr|
|-----|------:|----------------|-----:|-----------|---|-----:|---|-----:|
|gsm8k|      3|flexible-extract|     5|exact_match|↑  |0.6899|±  |0.0127|
|     |       |strict-match    |     5|exact_match|↑  |0.6384|±  |0.0132|

Us:

|Tasks|Version|     Filter     |n-shot|  Metric   |   |Value |   |Stderr|
|-----|------:|----------------|-----:|-----------|---|-----:|---|-----:|
|gsm8k|      3|flexible-extract|     5|exact_match|↑  |0.7036|±  |0.0126|
|     |       |strict-match    |     5|exact_match|↑  |0.6573|±  |0.0131|

[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 introduces a new lm-eval test case specifically designed for the NVFP4 quantization scheme, using a Llama 3.1 model and a specific dataset. It updates the existing lm-eval test infrastructure to automatically pick up this new configuration and also enhances test flexibility by making the number of calibration samples configurable via the test's YAML definition.

Highlights

• New Test Case: Added a new lm-eval test configuration (w4a4_nvfp4.yaml) to validate the NVFP4 quantization scheme with the meta-llama/Llama-3.1-8B-Instruct model and HuggingFaceH4/ultrachat_200k dataset, including expected exact_match metrics.
• Test Configuration Default: Updated test_lmeval.py to default to the newly added w4a4_nvfp4.yaml configuration file for lm-eval tests, ensuring the new test case is run by default.
• Configurable Calibration Samples: Modified the set_up method in test_lmeval.py to dynamically load num_calibration_samples from the lm-eval configuration, falling back to 512 if not specified, improving test flexibility.

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 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 or fill out our survey 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

This pull request adds a new lm-eval test case for NVFP4 quantization. The changes include a new YAML configuration file and modifications to the test script to load configuration values from it. I've suggested making the test configuration more explicit and avoiding potential KeyError exceptions.

[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.