Skip to content

[CI] Require transformers 5.19 for the EmbeddingGemma2 registry test - #60295

Merged
khluu merged 2 commits into
vllm-project:mainfrom
khluu:claude/embeddinggemma2-min-transformers
Oct 6, 2026
Merged

khluu merged 2 commits into
vllm-project:mainfrom
khluu:claude/embeddinggemma2-min-transformers

Conversation

@khluu

@khluu khluu commented Oct 6, 2026 •

Copy link
Copy Markdown
Member

Overview

Fix models/test_registry.py::test_registry_imports[EmbeddingGemma2Model], which has failed on main since #60254. The test registry entry now requires transformers 5.19.0, the first release that has transformers.models.embedding_gemma2.

Claims

Validation

Ran in the CI image for main bb87d227 (includes #60254, transformers==5.18.0):

pytest "models/test_registry.py::test_registry_imports[EmbeddingGemma2Model]"
Before 1 failed: ModuleNotFoundError: No module named 'transformers.models.embedding_gemma2'
After 1 skipped: "transformers==5.18.0 installed, but transformers>=5.19.0 is required to run this model."

pytest models/test_registry.py with the change: 392 passed, 17 skipped, 0 failed.

Details

vllm/model_executor/models/embedding_gemma2.py imports transformers.models.embedding_gemma2 at module level. That module is in transformers v5.19.0, released 2026-10-06, but not in v5.18.0. vLLM requires transformers >= 5.16.1, < 5.19.0 and CI pins 5.18.0. The registry entry said min_transformers_version="5.18.0", so check_transformers_version(on_fail="skip") passed, ModelRegistry._try_load_model_cls caught the import error and returned None, and the test failed on assert model_cls is not None.

The test will start running again once CI moves to transformers 5.19. The open follow-up #60289 doesn't touch tests/models/registry.py, so the two don't overlap. cc @lucianommartins (author of #60254)

This isn't a duplicate: I searched open PRs for EmbeddingGemma2 and test_registry_imports and found only #60289.


Pull Request Checklist
  • I used vLLM's /pr-checklist skill. (Mandatory for agents, optional for humans).

  • AI assistance was used during the creation of this PR.

  • Design Fit: Minimizes impact on core components, reuses existing functionality, and justifies added complexity.

  • Testing and Validation: Validates the change and ensures any added tests are meaningful and reliable, with CI coverage or documented CI resource constraints and validation performed outside CI.

  • Code Quality and Style: Keeps code and comments clear and concise, and updates relevant documentation and examples.

  • Pull Request Contents: Includes a brief summary and relevant links, supports claims with evidence, explains root causes and implementation trade-offs, and follows the contributing guide.

Claude Code wrote this change and ran the validation above. @khluu has not reviewed it yet; per AGENTS.md, he needs to review every line before it merges.

🤖 Generated with Claude Code

transformers.models.embedding_gemma2 first ships in transformers 5.19.0, but
the test registry entry added in vllm-project#60254 set min_transformers_version to
5.18.0. CI pins transformers==5.18.0, so test_registry_imports does not skip
and fails with ModuleNotFoundError, turning Basic Models (Other) red on
NVIDIA and AMD in every main build since.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: khluu <khluu000@gmail.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.

@AndreasKaratzas

Copy link
Copy Markdown
Member

/ci run

@github-actions

github-actions Bot commented Oct 6, 2026

Copy link
Copy Markdown

✅ Triggered Buildkite CI #93186 for commit d86db226e5b1.

@khluu
khluu merged commit ae19af8 into vllm-project:main Oct 6, 2026
39 of 41 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants