[Test] Add OmniGen2 online serving expansion L4 tests for Ulysses-SP and CFG-Parallel

[yangjianjuan]

Purpose

This PR adds L4 online serving expansion tests for OmniGen2 models
test features:

• Ulysses-SP
• CFG-Parallel

Test Plan

pytest test_omni_gen2_expansion.py -m 'diffusion' -s -v

Test Result

<frozen importlib._bootstrap>:488
  <frozen importlib._bootstrap>:488: DeprecationWarning: builtin type SwigPyPacked has no __module__ attribute

<frozen importlib._bootstrap>:488
  <frozen importlib._bootstrap>:488: DeprecationWarning: builtin type SwigPyObject has no __module__ attribute

../../../../../.venv/lib/python3.12/site-packages/torch/jit/_script.py:362: 14 warnings
  /home/.venv/lib/python3.12/site-packages/torch/jit/_script.py:362: DeprecationWarning: `torch.jit.script_method` is deprecated. Please switch to `torch.compile` or `torch.export`.
    warnings.warn(

../../../vllm_omni/__init__.py:29
  /home/y00623537/vllm-omni/vllm_omni/__init__.py:29: RuntimeWarning: Failed to import version from _version.py: No module named 'vllm_omni._version'
  This typically happens in development mode before building.
  Using fallback version 'dev'.
    from .version import __version__, __version_tuple__  # isort:skip

../../../../../.venv/lib/python3.12/site-packages/_pytest/config/__init__.py:1428
  /home/.venv/lib/python3.12/site-packages/_pytest/config/__init__.py:1428: PytestConfigWarning: Unknown config option: asyncio_mode
  
    self._warn_or_fail_if_strict(f"Unknown config option: {key}\n")

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
============================================================================================ 2 passed, 18 warnings in 315.21s (0:05:15) ============================================================================================

---

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan. Please provide the test scripts & test commands. Please state the reasons if your codes don't require additional test scripts. For test file guidelines, please check the test style doc
• The test results. Please paste the results comparison before and after, or the e2e results.
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model. Please run mkdocs serve to sync the documentation editions to ./docs.
• (Optional) Release notes update. If your change is user-facing, please update the release notes draft.

BEFORE SUBMITTING, PLEASE READ https://github.com/vllm-project/vllm-omni/blob/main/CONTRIBUTING.md (anything written below this line will be removed by GitHub Actions)

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: be6ce66aeb

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Drop CFG-parallel case for OmniGen2 expansion test

This case claims CFG-parallel coverage for OmniGen2, but OmniGen2Pipeline does not use CFGParallelMixin (or any rank-aware CFG split path), so --cfg-parallel-size 2 does not actually exercise CFG-parallel behavior and can pass while that feature is effectively untested for this model. In practice this creates false confidence: regressions in real CFG-parallel execution would not be caught by this test.

Useful? React with 👍 / 👎.

[wtomin]

Could you double check it? Cause I think OmniGen2 does not support SP or CFG-Parallel for now.

Besides, there is RFC related to L4 tests #1832 . Please request for task assignment first.

[yangjianjuan]

Could you double check it? Cause I think OmniGen2 does not support SP or CFG-Parallel for now.

Besides, there is RFC related to L4 tests #1832 . Please request for task assignment first.

OK :-)

[wtomin]

I double checked, and realized that if usp and cfg-parallel were enabled, but those models didn't support the two features, the inference will not fail, but disable the two features automatically.

Sorry I will postpone the merge of this PR, until OmniGen2 indeed support the two features.

Would you like to raise a PR to support SP for OmniGen2? I will sync it to #1217 . CFG-Parallel of OmniGen2 is in #2074.

[yangjianjuan]

I double checked, and realized that if usp and cfg-parallel were enabled, but those models didn't support the two features, the inference will not fail, but disable the two features automatically.

Sorry I will postpone the merge of this PR, until OmniGen2 indeed support the two features.

Would you like to raise a PR to support SP for OmniGen2? I will sync it to #1217 . CFG-Parallel of OmniGen2 is in #2074.

OK, I accept to support SP for OmniGen2.

[wtomin]

@yangjianjuan Hi, may I ask about the progress of the SP support for OmniGen2?

[yangjianjuan]

@yangjianjuan Hi, may I ask about the progress of the SP support for OmniGen2?

Sorry, I'm currently focused on another priority. I’ll be able to get started on this next week.