[CPU][CI] Fallback sliding window to v0 and fix CPU pooling model tests

[bigPYJ1151]

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan, such as providing test command.
• The test results, such as pasting the results comparison before and after, or e2e results
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.

Purpose

• Fallback models with sliding window to V0 as CPU chunked prefill attention kernel doesn't support sliding window.
• Fix CPU embedding model tests.

Test Plan

Test Result

(Optional) Documentation Update

[gemini-code-assist]

Summary of Changes

Hello @bigPYJ1151, 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 primarily addresses compatibility issues with models utilizing sliding window attention on CPU, ensuring proper fallback to the V0 engine where the V1 CPU kernel lacks support. Additionally, it includes targeted fixes to improve the reliability and accuracy of CPU embedding model tests by accounting for these architectural limitations.

Highlights

• CPU Sliding Window Handling: Implemented a mechanism within vllm/engine/arg_utils.py to prevent the V1 engine from being used for models that utilize sliding window attention when running on a CPU backend. This ensures a fallback to the V0 engine, as the current CPU chunked prefill attention kernel in V1 does not support sliding window.
• CPU Embedding Model Test Fixes: Adjusted tests/models/language/pooling/test_embedding.py to correctly mark the ssmits/Qwen2-7B-Instruct-embed-base model for CPU-specific testing. Additionally, a pytest.skip condition was added for the intfloat/e5-mistral-7b-instruct model when running on CPU with V1 enabled, explicitly skipping tests for this configuration due to the lack of sliding window support in CPU V1.

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. ↩

[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]

Code Review

This pull request aims to address compatibility issues for models using sliding window attention when running on CPU, specifically by falling back to the V0 engine, and also includes fixes for CPU pooling model tests.

The code changes are well-targeted and appear correct:

1. In vllm/engine/arg_utils.py, the modification to _is_v1_supported_oracle correctly adds logic to detect if a model with a sliding window is running on a CPU, and if so, ensures it falls back to the V0 engine. This directly implements the primary goal of the PR.
2. In tests/models/language/pooling/test_embedding.py:
   • The test for intfloat/e5-mistral-7b-instruct is now skipped when running on CPU with the V1 engine, which is appropriate given that CPU V1 doesn't support sliding window for this model.
   • The ssmits/Qwen2-7B-Instruct-embed-base test parameterization now includes pytest.mark.cpu_model, aligning with the goal of fixing/improving CPU model tests.
   • The import of the os module is necessary for the environment variable check.

The changes are clear and directly support the PR's objectives. No medium, high, or critical issues were identified in the provided code diffs.

To further improve the pull request quality, I recommend completing the PR description by filling out the "Test Plan" and "Test Result" sections. This will help reviewers and future maintainers understand how these changes were validated.