Skip to content

[CI] Fix MiniMax-M3 PP aux-state test mock after #58648 - #59508

Merged
njhill merged 1 commit into
vllm-project:mainfrom
venkywonka:fix/eagle3-pp-test-mock-58648
Oct 1, 2026
Merged

njhill merged 1 commit into
vllm-project:mainfrom
venkywonka:fix/eagle3-pp-test-mock-58648

Conversation

@venkywonka

Copy link
Copy Markdown
Contributor

Purpose

tests/v1/worker/test_eagle3_aux_hidden_states_pp.py fails on main with 5× AttributeError: 'types.SimpleNamespace' object has no attribute 'use_v2_model_runner' (seen in the v1-executor-plus-worker step on NVIDIA and AMD).

#58648 made MiniMaxM3Model.__init__ read vllm_config.use_v2_model_runner and vllm_config.speculative_config. #57197 added this test with a SimpleNamespace config that has neither, and it merged on CI results from before #58648. Each PR passed on its own; together they break.

This adds speculative_config=None and use_v2_model_runner=False to the test's mock config. With no speculative config, spec_decode_needs_target_embed returns False, so the test exercises the same behavior as before #58648. Test-only change, +4/−1.

Not a duplicate: no open PR or issue references this test or failure.

Test Plan

pytest tests/v1/worker/test_eagle3_aux_hidden_states_pp.py \
       tests/v1/worker/test_spec_decode_embed_sharing_pp.py

Test Result

On an RTX PRO 6000 (SM120):

revision passed failed
main 37 5 (the use_v2_model_runner AttributeError)
this PR 42 0
  • Submitter reviewed every changed line

AI assistance (Claude Code) was used to find the cause, write the fix and run the tests.

🤖 Generated with Claude Code

vllm-project#58648 made MiniMaxM3Model read vllm_config.use_v2_model_runner and
speculative_config. The test's SimpleNamespace config lacked both, so
test_eagle3_aux_hidden_states_pp.py fails with AttributeError on main.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Signed-off-by: venkywonka <23023424+venkywonka@users.noreply.github.com>

@claude claude Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Claude Code Review

This pull request is from a fork — automated review is disabled. A repository maintainer can comment @claude review to run a one-time review.

@njhill njhill added the ready ONLY add when PR is ready to merge/full CI is needed label Sep 30, 2026
@njhill

njhill commented Sep 30, 2026

Copy link
Copy Markdown
Member

/ci run

@github-actions

Copy link
Copy Markdown

✅ @venkywonka, CI is now available for this PR.

  • /ci run starts upstream CI; /amd-ci run starts AMD CI only.
  • Your branch must contain every commit currently on its upstream target branch. Merge or rebase onto the latest target branch, then rerun the command. Append --allow-stale to a run command to test an outdated branch at your own risk.
  • /ci retry retries failed jobs in the CI build for the current PR head. If the current head has no CI build, it starts a new CI build for the current head containing only jobs that failed in the latest earlier CI build for this PR.
  • /amd-ci retry retries failed jobs in AMD CI for the current PR head. Use /amd-ci run when the current head has no AMD CI build.
  • /ci cancel cancels scheduled or running CI builds for this PR branch; /amd-ci cancel does the same for AMD CI only.

@github-actions

Copy link
Copy Markdown

✅ Triggered Buildkite CI #92277 for commit 2fe7a6021592.

@njhill
njhill merged commit d7c5511 into vllm-project:main Oct 1, 2026
46 of 48 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

minimax ready ONLY add when PR is ready to merge/full CI is needed

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants