Install tpu_info==0.4.0 to fix core dump for TPU #23135
Conversation
|
👋 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 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 🚀 |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request pins the tpu_info package to version 0.4.0 to fix a core dump issue found in a newer version. This is a sensible and direct fix for the CI problem. My review includes one suggestion to improve the maintainability of this change by adding a reference to a tracking issue.
xiangxu-google
force-pushed
the
fix_tpu_info
branch
from
a9e2dfa to
288f07b
Compare
yaochengji
left a comment
yaochengji
left a comment
There was a problem hiding this comment.
LGTM, thanks for the quick fix!
Signed-off-by: Duncan Moss <djm.moss@gmail.com>
Signed-off-by: Xiao Yu <xiao.yu@amd.com>
3 participants
Purpose
To fix the broken TPU CI https://buildkite.com/vllm/ci/builds/27465#0198be9b-3af1-4abf-929c-c66d8fcebc95
Test Plan
Local test
pytest -s -v tests/v1/tpu/test_perf.pyTest Result
Passed
(Optional) Documentation Update
Essential Elements of an Effective PR Description Checklist
supported_models.mdandexamplesfor a new model.