Disable block verify to avoid incorrect verification on NPU

[liuchenbing2026]

What this PR does / why we need it?

Does this PR introduce any user-facing change?

How was this patch tested?

• vLLM version: v0.18.0
• vLLM main: vllm-project/vllm@ed359c4

[gemini-code-assist]

Summary of Changes

Hello, 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 implements a critical fix to prevent incorrect block verification on NPU devices. By unconditionally disabling the using_block_verify flag within the rejection sampling logic, the change ensures stable and accurate operation when running on NPUs, mitigating potential verification errors that could arise from the previous conditional behavior.

Highlights

• Block Verification Disabled: The using_block_verify flag in vllm_ascend/sample/rejection_sampler.py has been explicitly set to False, overriding its previous conditional assignment based on max_spec_len.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

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.

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 disables the block verification feature in the rejection sampler by hardcoding using_block_verify to False. This is a workaround for a bug causing incorrect verification on NPUs. My review includes a suggestion to update a related code comment for clarity and maintainability. I have also provided a suggested PR title and summary to align with the repository's style guide.

Suggested PR Title:

[Sample][BugFix] Disable block verify to avoid incorrect verification on NPU

Suggested PR Summary:

### What this PR does / why we need it?

This PR disables the "block verify" feature within the rejection sampler for speculative decoding on Ascend NPUs. This is achieved by setting the `using_block_verify` flag to `False`.

This change is a workaround for a bug causing incorrect verification results when using the block verification logic on NPUs, which could lead to correctness issues. Disabling this feature ensures correct behavior.

### Does this PR introduce _any_ user-facing change?

No. This is an internal bug fix and is not expected to have any user-facing impact, aside from potential performance changes in speculative decoding.

### How was this patch tested?

CI is expected to pass. This change disables a feature to prevent a bug. Further testing should validate that the bug is resolved and that the performance impact is acceptable.

[gemini-code-assist]

Hardcoding using_block_verify to False disables the feature, but the comment on the preceding line is now misleading. To improve maintainability, the comment on line 113 should be updated to explain why block verification is disabled and include a TODO to track re-enabling it once the underlying NPU bug is fixed. For example:

# Block verify is disabled to avoid incorrect verification on NPU.
# TODO: Re-enable block verify when the bug is fixed.

[github-actions]

👋 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.