update deepep version for qwen3-next deepep moe#10624
Conversation
yizhang2077
requested review from
ByronHsu,
HaiShaw and
zhyncs
as code owners
![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 @yizhang2077, 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 focuses on updating the DeepEP library dependency to a newer version. The core purpose of this update is to integrate crucial changes from DeepEP that facilitate support for Qwen3-Next models, particularly those configured with 512 experts, thereby enhancing the system's compatibility and capabilities with advanced model architectures.
Highlights
- DeepEP Version Update: The commit hash for the DeepEP library has been updated from
b92d0d4860ce6866cd6d31bfbae937f9a7a3772bto9af0e0d0e74f3577af1979c9b9e1ac2cad0104eein both the Dockerfile and the CI installation script. - Qwen3-Next Support: This update specifically enables support for Qwen3-Next models with 512 experts, leveraging recent changes introduced in the DeepEP repository.
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
-
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 updates the DeepEP dependency to a newer commit to support qwen3-next models with 512 experts. The change is applied consistently in both the Dockerfile and the CI installation script. My review focuses on improving maintainability by centralizing the dependency version to avoid duplication.
| ARG BUILD_TYPE=all | ||
| ARG BRANCH_TYPE=remote | ||
| ARG DEEPEP_COMMIT=b92d0d4860ce6866cd6d31bfbae937f9a7a3772b | ||
| ARG DEEPEP_COMMIT=9af0e0d0e74f3577af1979c9b9e1ac2cad0104ee |
There was a problem hiding this comment.
This DEEPEP_COMMIT value is also hardcoded in scripts/ci/ci_install_deepep.sh. To improve maintainability and prevent potential inconsistencies during future updates, it would be beneficial to establish a single source of truth for this commit hash. For instance, you could define it in a CI/CD environment variable and pass it as a --build-arg to Docker and export it for the shell script.
|
|
||
| # Install DeepEP | ||
| rm -rf /root/.cache/deepep && git clone https://github.com/deepseek-ai/DeepEP.git /root/.cache/deepep && cd /root/.cache/deepep && git checkout b92d0d4860ce6866cd6d31bfbae937f9a7a3772b | ||
| rm -rf /root/.cache/deepep && git clone https://github.com/deepseek-ai/DeepEP.git /root/.cache/deepep && cd /root/.cache/deepep && git checkout 9af0e0d0e74f3577af1979c9b9e1ac2cad0104ee |
There was a problem hiding this comment.
The DeepEP commit hash is hardcoded here, and it's also present in docker/Dockerfile. This duplication makes future updates more error-prone. To improve maintainability, consider using an environment variable. This allows you to set the commit hash from a single source of truth in your CI environment, with a fallback to the hardcoded value.
| rm -rf /root/.cache/deepep && git clone https://github.com/deepseek-ai/DeepEP.git /root/.cache/deepep && cd /root/.cache/deepep && git checkout 9af0e0d0e74f3577af1979c9b9e1ac2cad0104ee | |
| rm -rf /root/.cache/deepep && git clone https://github.com/deepseek-ai/DeepEP.git /root/.cache/deepep && cd /root/.cache/deepep && git checkout ${DEEPEP_COMMIT:-9af0e0d0e74f3577af1979c9b9e1ac2cad0104ee} |
* origin/qwen3: (30 commits) chore: bump sgl-kernel 0.3.11 (sgl-project#10630) feat: add fused moe config for Qwen3-Next-80B-A3B-Instruct on B200 (sgl-project#10631) model support: Sarashina2VisionForCausalLM (sgl-project#10632) [Performance] Qwen3-Next: speed up update_mamba_state_after_mtp_verify by 10x; e2e up to 3.54% faster (sgl-project#10586) [Performance] Qwen3-Next: replace arange to cached query_start_loc_li… (sgl-project#10553) [Feature] Speculative decoding support lookahead (sgl-project#9873) refactor: use registry for _get_attention_backend_from_str (sgl-project#10629) [router] refactor worker to builder pattern 1/n (sgl-project#10628) Garbage collector regression in the online server (sgl-project#10621) feat: Add FlexAttention Backend for Efficient Sparse Attention (sgl-project#9947) Fix bias handling in TritonMoeQuantInfo within quantization/mxfp4.py (sgl-project#10579) [Performance] qwen3-next improve causal conv1d in prefill phase (sgl-project#10595) Fix sgl_kernel import failure on devices other than CUDA (sgl-project#10610) support qwen3-next-fp8 deepep (sgl-project#10622) update deepep version for qwen3-next deepep moe (sgl-project#10624) Feat/add heartbeat mechanism for nixl conn (sgl-project#10222) [RL] Add destroy process group api (sgl-project#9979) fix deepep assert when PD disaggregation == null (sgl-project#8274) Scale kkt after reduction (sgl-project#10604) [improvement] add average input/output token length for hicache benchmark stats output (sgl-project#10525) ...
3 participants
Motivation
Support qwen3-next expert number = 512 since deepseek-ai/DeepEP#403
Modifications
Accuracy Tests
Benchmarking and Profiling
Checklist