Issue Station
Sign In
Tools hereToolbox

pytorch - ✅(Solved) Fix [release 2.11][vllm] PyTorch Compilation Unit tests failure [1 pull requests, 6 comments, 2 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
pytorch/pytorch#176833•Fetched 2026-04-08 00:24:14
View on GitHub
Comments
6
Participants
2
Timeline
35
Reactions
0
Author
zou3519zou3519
Participants
zhxchen17zhxchen17zou3519zou3519
Assignees
zhxchen17zhxchen17
Timeline (top)
mentioned ×7subscribed ×7commented ×6labeled ×6

Fix Action

Fixed

PR fix notes

PR #36582: [compile] Apply stored functorch config while finalizing loaded artifacts.

Description (problem / solution / changelog)

Summary:

Fixing test failure while updating to torch 2.11 RC from torch 2.10.

After recent refactoring on logging, we didn't apply serialized functorch configs again causing aotautograd config change and cache miss in warm start. This PR fixes the issue by re-applying these stored configs in the new path.

Context: https://github.com/pytorch/pytorch/issues/176833

Test Plan: pytest tests/compile/test_startup.py

passing with torch 2.10, 2.11RC, 2.12trunk

Reviewers:

Subscribers:

Tasks:

Tags:

Signed-off-by: zhxchen17 [email protected]

Purpose

Test Plan

Test Result

<details> <summary> Essential Elements of an Effective PR Description Checklist </summary>
  • 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.
  • (Optional) Release notes update. If your change is user facing, please update the release notes draft in the Google Doc.
</details>

Changed files

  • vllm/compilation/caching.py (modified, +7/-1)
  • vllm/compilation/piecewise_backend.py (modified, +8/-24)
RAW_BUFFERClick to expand / collapse

[2026-03-05T15:16:07Z] =========================== short test summary info ============================

[2026-03-05T15:16:07Z] FAILED compile/test_startup.py::test_moe_startup - AssertionError: num_compiled_artifacts_saved not as expected, before it is 0, after it is 1, expected diff is 0

https://buildkite.com/vllm/ci/builds/54647#019cbd53-c664-4f76-a0c5-94de7c35349d

cc @ezyang @gchanan @kadeng @msaroufim @chauhang @penguinwu

extent analysis

Fix Plan

Fix Name: Update Test Expectations

Step-by-Step Solution

1. Identify the affected test

Update the test_moe_startup test in compile/test_startup.py to reflect the correct expected value for num_compiled_artifacts_saved.

2. Update test expectations

Change the assertion in the test_moe_startup test to expect a difference of 1 instead of 0.

# Before (incorrect)
self.assertEqual(num_compiled_artifacts_saved, 0)

# After (correct)
self.assertEqual(num_compiled_artifacts_saved, 1)

3. Verify the fix

Run the test again to ensure it passes with the updated expectations.

4. Review and commit changes

Review the changes and commit them to the repository with a descriptive commit message.

git add compile/test_startup.py
git commit -m "Update test expectations for test_moe_startup"

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#api#ssr#installation#tensor shape#autograd error#indexing error#inference speed#output truncation#response parsing#generation error

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING