[WIP] [MoE] GPT OSS #1705
[WIP] [MoE] GPT OSS #1705
Conversation
|
👋 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[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
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 initial support for GPT-OSS (Open-Source Software) models, specifically focusing on their Mixture-of-Experts (MoE) components, within the llmcompressor framework. It includes a new module to handle individual experts and a linear layer representation for expert groups, along with preparatory steps for integrating these models into the compression pipeline. The "WIP" status indicates ongoing development.
Highlights
- GPT-OSS Expert Implementation: I've added GptOssExpert and GptOssExpertsLinear classes in src/llmcompressor/modeling/gpt_oss.py. These classes are designed to re-implement or wrap the expert layers from GPT-OSS models, allowing for potential optimization or specific handling within llmcompressor.
- MoE Integration Preparation: I've updated src/llmcompressor/modeling/prepare.py by adding a placeholder update_gpt_oss_moe function and registering it in the moe_context. This sets the stage for integrating GPT-OSS MoE models into the model preparation and compression pipeline.
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
-
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. ↩
There was a problem hiding this comment.
Code Review
This pull request adds initial support for GPT-OSS models, including a new GptOssExpertsLinear module and scaffolding for model preparation. The changes are a good starting point but are clearly a work-in-progress. My review focuses on improving the correctness and clarity of the new PyTorch module. Key feedback includes using torch.nn.ModuleList for proper submodule registration, correcting type hints and function signatures for better API design, and addressing an unimplemented function and a leftover breakpoint(). I've also noted that the new preparation function for GPT-OSS needs to be registered to be effective.
|
awesome!! I was just looking at this to convert to FP8, since MXFP4 will not work on ADA Series |
Signed-off-by: Kyle Sayers <[email protected]>
kylesayrs
force-pushed
the
kylesayrs/gpt_oss
branch
from
6d55c1a to
affa73f
Compare
Signed-off-by: Kyle Sayers <[email protected]>
Signed-off-by: Kyle Sayers <[email protected]>
Purpose
Changes