[wip] Support Transformers v5.x

[byjiang1996]

Motivation

From #14356, it is required for supporting glm 4.6v

Co-authored-by: @yhyang201

Modifications

• pip install required packages and bump package versions
• alwasy rebuild sgl-kernel's flash_attn even if flash_attn exists in the pip: Rebuilding from flash_attn main branch is required for fa4 as fa4 is not yet released into pip dist.

Accuracy Tests

Benchmarking and Profiling

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.
• Work with maintainers to merge your PR. See the PR Merge Process

[byjiang1996]

/tag-and-rerun-ci

[gemini-code-assist]

Summary of Changes

Hello @byjiang1996, 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 updates the project's dependency stack to support the upcoming Transformers v5.x, which is crucial for integrating with newer models like GLM 4.6v. The changes involve upgrading the transformers package to its release candidate version and introducing flash_attn as a direct dependency. To ensure compatibility with the latest Flash Attention features, specifically FA4, the build process for sgl-kernel has been adjusted to force a fresh installation of flash_attn from its main branch. Additionally, some internal configuration validation logic related to rope_config_validation has been removed to align with the updated Transformers API.

Highlights

• Dependency Updates: Upgraded the transformers library to 5.0.0rc0 across all pyproject.toml files and added flash_attn as a new dependency.
• Flash Attention Build Process: Modified the sgl-kernel build system to ensure a forced rebuild and proper installation of flash_attn from its main branch, specifically to support FA4 which is not yet released via pip.
• Transformers API Adaptation: Removed references and calls to rope_config_validation in qwen3_omni.py, adapting to potential API changes in Transformers v5.x.

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 project to support transformers v5.x by bumping the version to 5.0.0rc0 and adding flash_attn as a dependency. The changes also include necessary adaptations in the codebase to accommodate breaking changes in the new transformers version, such as the removal of rope_config_validation. Additionally, the build process for sgl-kernel is modified to ensure flash_attn is built from source to include support for fa4.

My feedback focuses on improving dependency management for better reproducibility. While using a release candidate for transformers is a conscious choice for this upgrade, it's important to be aware of potential stability risks. The changes to the build system are well-justified workarounds.

[gemini-code-assist]

It's a good practice to pin dependency versions to ensure reproducible builds. The flash_attn dependency is added without a version specifier, which could lead to unexpected issues if a new, incompatible version is released. Please consider pinning it to a specific version (e.g., flash_attn=="x.y.z") or a minimum compatible version (e.g., flash_attn>="x.y.z").