[model] support Qwen/Qwen3-VL-30B-A3B-Instruct/Thinking

[Jintao-Huang]

No description provided.

[gemini-code-assist]

Summary of Changes

Hello @Jintao-Huang, 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 significantly enhances the system's model capabilities by integrating several new Qwen3-VL models, including their instruction-tuned, 'thinking,' and FP8 optimized variants. This ensures that users have access to the latest Qwen models and their efficient versions for various tasks. The changes also include necessary updates to the documentation to reflect these additions and a minor cleanup in the trainer module's import structure.

Highlights

• New Model Support: Added support for Qwen3-VL-30B-A3B-Instruct and Qwen3-VL-30B-A3B-Thinking models, expanding the range of available Qwen models.
• FP8 Model Integration: Included support for FP8 optimized versions of Qwen3-VL-30B and Qwen3-VL-235B models, allowing for more efficient inference.
• Documentation Update: Updated both Chinese and English documentation files to accurately reflect the newly supported Qwen3-VL models and their variants.
• Trainer Module Refactoring: Performed a minor refactoring in the trainer module by removing EvaluationStrategy from import paths in swift/__init__.py and swift/trainers/__init__.py.

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 adds support for new Qwen3-VL models, including 30B and 235B variants with FP8 quantization. The changes are well-organized, with consistent updates to the model registration in swift/llm/model/model/qwen.py and the corresponding documentation in both Chinese and English. The code also includes a minor but beneficial cleanup by removing the deprecated EvaluationStrategy import from the trainers module. Overall, the changes are correct and improve the project. Good work!

[Jintao-Huang]

/gemini review

[gemini-code-assist]

Code Review

This pull request adds support for new Qwen3-VL models and updates compatibility with newer versions of the transformers library. The changes are generally well-implemented, including necessary patches for model compatibility and forward-compatible adjustments for the trainer. I've identified one medium-severity issue where the code for model registration doesn't match the documentation regarding Megatron support, which could lead to confusion or unexpected behavior for users.

[mertunsall]

Hi @Jintao-Huang - could you please share with me the number of H100s for full parameter fine-tune of Qwen3-VL-30B-A3B? Should 2 nodes of 8xH100 suffice? or maybe 8xB200?

I will use Megatron