Skip to content

Conversation

@hengtaoguo
Copy link
Collaborator

@hengtaoguo hengtaoguo commented Jan 29, 2026

Description

Decouple multimodal_utils.py into centralized router and model-specific functions.

  • multimodal/utils.py: Commonly shared basic functions and constant.
  • multimodal/processor.py: Centralized router. Routes and manages multimodal data preprocessing, prompt formatting, and input preparation for different model types by re-directing to model-specific processor modules.
  • Updated dependencies for models.py/decoders.py/maxengine.py etc for the above changes.

Tests

Unit tests:

python -m unittest tests.unit.multimodal_utils_test
python -m unittest tests.unit.qwen3_omni_layers_test.TestQwen3OmniPreprocessing

End2end tests:

SFT tests:

Checklist

Before submitting this PR, please make sure (put X in square brackets):

  • I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
  • I have necessary comments in my code, particularly in hard-to-understand areas.
  • I have run end-to-end tests tests and provided workload links above if applicable.
  • I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.

@hengtaoguo hengtaoguo force-pushed the hengtaoguo-utils branch 3 times, most recently from bdb4730 to f345767 Compare January 30, 2026 00:33
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.

1 participant