👋 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.

🚀

Looks like xgrammar

(.venv) sage:~/git/vllm> pip show triton
Name: triton
Version: 3.3.0
...
Requires: setuptools
Required-by: xgrammar

I suspect it is just pulling in the triton package without specifying the version. I tried this on an nvidia system and I think we dodge the issue because torch explicitly depends on the triton package instead of bundling it

(.venv) sage@beaker ~/g/nm-vllm (main)> pip list | grep triton
triton                                   3.2.0
(.venv) sage@beaker ~/g/nm-vllm (main)> pip show triton
Name: triton
Version: 3.2.0
Summary: A language and compiler for custom Deep Learning operations
Home-page: https://github.com/triton-lang/triton/
Author: Philippe Tillet
Author-email: phil@openai.com
License:
Location: /home/sage/git/nm-vllm/.venv/lib/python3.10/site-packages
Requires:
Required-by: torch, xgrammar