Skip to content

Conversation

@shenchuxiaofugui
Copy link
Contributor

@shenchuxiaofugui shenchuxiaofugui commented Nov 29, 2025

What this PR does / why we need it?

When the dtype of the MTP layer is fp16, this layer does not go through EPLB. However, the EPLB check in the verification process is applied uniformly to the entire model, and it still checks whether the dtype is w8a8. Currently, whether to use EPLB is determined separately for each layer.

Does this PR introduce any user-facing change?

No

How was this patch tested?

@github-actions
Copy link

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

  • A PR should do only one thing, smaller PRs enable faster reviews.
  • Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
  • Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

Signed-off-by: shenchuxiaofugui <1311027364@qq.com>
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request addresses a bug in the Expert Parallel Load Balancing (EPLB) verification. The change correctly restricts the quantization method check to static EPLB configurations (those using expert_map_path), resolving an issue where dynamic EPLB was incorrectly constrained. The implementation is sound and the fix is effective. The changes look good.

Signed-off-by: shenchuxiaofugui <1311027364@qq.com>
Signed-off-by: shenchuxiaofugui <1311027364@qq.com>
@wangxiyuan wangxiyuan closed this Dec 2, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants