Skip to content

Conversation

tianyuan211
Copy link

@tianyuan211 tianyuan211 commented Sep 28, 2025

Summary

This PR adds support for integrating OVIS2.5 into vLLM for improved inference performance.

Changes

  • Integrated OVIS2.5 as an inference backend.
  • Updated configuration files for OVIS2.5 support.
  • Added documentation for setup and usage.

Motivation

The integration improves inference speed and scalability for large language models.

Test

python3 examples/offline_inference/vision_language.py -m 'ovis2_5'

Signed-off-by: tianyuan211 <[email protected]>
@czhu15
Copy link

czhu15 commented Sep 30, 2025

Suggest to add an example code on how to run this model on the description section.

)
messages = [{"role": "user", "content": f"{placeholders}\n{question}"}]

tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
Copy link

Choose a reason for hiding this comment

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

should we use "AIDC-AI/Ovis2-1B" tokenizer here too?

# python launch_multi_models.py --models model1 model2 --max-model-len 4096
# Performance preset
# python launch_multi_models.py --models model1 model2 --env-preset \
# performance
Copy link

Choose a reason for hiding this comment

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

why change "launch_multi_models.py" file?

"""Load environment variables from JSON config file."""
try:
with open(config_path, 'r') as f:
with open(config_path) as f:
Copy link

Choose a reason for hiding this comment

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

remove 'r' here seems not correct.

@tianyuan211
Copy link
Author

Suggest to add an example code on how to run this model on the description section.

fixed

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