GLM-4 Update

[zRzRzRzRzRzRzR]

GLM-4 Update of Moe Support

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

[gemini-code-assist]

Summary of Changes

Hello @zRzRzRzRzRzRzR, 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 comprehensive support for the GLM-4 Mixture-of-Experts (MoE) model within vLLM. It provides the full architectural implementation, including specialized MoE components, and integrates the new model into the existing benchmarking and model loading infrastructure.

Highlights

• New Model Support: I've added the complete implementation for the Glm4MoeForCausalLM model architecture, enabling vLLM to support GLM-4 Mixture-of-Experts models.
• Mixture-of-Experts (MoE) Implementation: The new glm4_moe.py file includes custom components like Glm4MoeTopkRouter for expert selection and Glm4MoeSparseMoeBlock to handle the specific routing and shared expert mechanisms of GLM-4's MoE architecture.
• System Integration: Updates were made to the MoE benchmarking script (benchmark_moe.py) and the model registry (registry.py) to properly recognize and integrate the new Glm4MoeForCausalLM model within vLLM's ecosystem.
• Weight Loading Logic: The new model implementation includes robust weight loading logic to correctly map and load parameters from HuggingFace checkpoints, covering stacked linear layers (QKV, gate/up projections) and individual expert weights.

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 is currently in preview and 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 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

The pull request adds support for GLM-4 MoE models to vLLM. The changes include modifying the benchmark script to include GLM-4 MoE, adding the GLM-4 MoE model to the model registry, and creating a new file for the GLM-4 MoE model architecture. There's a potential typo in the benchmark script that needs to be addressed, and the model path in the registry should be a real model name.

[jeejeelee]

Thank you, we also need to update the document.

[zRzRzRzRzRzRzR]

Thank you, we also need to update the document.

Thank you, we also need to update the document.

you mean supported_models.md, isnt it generate auto?

[jeejeelee]

Thank you, we also need to update the document.

Thank you, we also need to update the document.

you mean supported_models.md, isnt it generate auto?

Yes, we need to update manually.

[zRzRzRzRzRzRzR]

We also have the MTP model, but if this PR meets the requirements, it can be merged first. The implementation of MTP can be proposed in a separate PR.