[Examples] Qwen3_vl

[kylesayrs]

Purpose

• Fixes [Bug]: Qwen3-VL-8B-Instruct AWQ can't be done on GPU #2068
  • Offloading issue was fixed by [Offload] Switch to torch offloader from accelerate #2148

[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 introduces a new example for the Qwen3-VL-8B-Instruct multimodal model, demonstrating the complete workflow for applying AWQ quantization. It covers loading the model, preparing a multimodal dataset, configuring the AWQ recipe, executing the oneshot compression, and verifying the quantized model's performance through sample generation. This addition is crucial for showcasing multimodal model compression capabilities and resolves a prior offloading issue, enhancing the robustness of the compression framework.

Highlights

• New Qwen3-VL Example: Introduces a comprehensive example script for applying AWQ quantization to the Qwen3-VL-8B-Instruct multimodal model, demonstrating the full compression workflow.
• AWQ Quantization Workflow: The example showcases the end-to-end process, including model loading, multimodal dataset preparation (Flickr30k), defining an AWQ recipe, performing oneshot quantization, and verifying the quantized model's generation capabilities.
• Multimodal Data Handling: Includes custom preprocessing for image data (base64 encoding) and a specific data collator designed for multimodal inputs, ensuring proper handling of diverse data types.
• Offloading Issue Resolution: This example is enabled by a recent fix for an offloading issue, ensuring correct and robust functionality for multimodal model compression within the framework.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

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

The pull request introduces a new example script for quantizing the Qwen3-VL model using the AWQModifier. The script demonstrates loading the model, preparing a multimodal dataset, applying the AWQ quantization recipe, and then performing sample generation and saving the compressed model. The overall structure is clear and follows the typical workflow for llmcompressor examples. One area for improvement is the robustness of the data collator regarding batch size expectations.

[kylesayrs]

@dsikka Do you know why mergify thinks that this is documentation?

[dsikka]

@dsikka Do you know why mergify thinks that this is documentation?

Any example file is technically a doc but we can change that rule

[brian-dellabetta]

🥇