[CI] Turn on basic correctness tests for V1

[tlrmchlsmth]

Ran into some OOM issues in #9856. Turning on this test to see if the problems happen without the TP changes.

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.
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 do one of these:

• Add ready label to the PR
• Enable auto-merge.

🚀

[WoosukKwon]

@tlrmchlsmth do we still have any issues on merging this PR?

[tlrmchlsmth]

@WoosukKwon I tried it again after merging #9856 but it's still red. Maybe there are issues with the UniprocExecutor that can be resolved one way or another ✂️

[wubai]

vllm/vllm/v1/worker/gpu_model_runner.py

Line 704 in c2d1b07

logits = self.model.compute_logits(hidden_states, None)

@tlrmchlsmth logits = self.model.compute_logits(hidden_states, None) comsume much more memory than V0, that lead to the OOM. But I don't know how to figure it out.

[tlrmchlsmth]

@tlrmchlsmth logits = self.model.compute_logits(hidden_states, None) comsume much more memory than V0, that lead to the OOM. But I don't know how to figure it out.

Hey, I'm pretty sure the failure was due to the engine cleaning up improperly previously, but curious what you were seeing before?

[mgoin]

Let's get this enabled