wangfuyun

/

Diffusion-NPO

like 0

This PR adds a comprehensive model card for Self-NPO, a data-free negative preference optimization approach, which is part of the broader Diffusion-NPO framework.

The update includes:

• A link to the paper: Self-NPO: Data-Free Diffusion Model Enhancement via Truncated Diffusion Fine-Tuning.
• Relevant metadata: license (Apache 2.0), pipeline_tag (text-to-image), and library_name (diffusers).
• The full abstract of the paper.
• An overview and key features of the model, extracted from the GitHub README.
• A detailed sample usage section for inference, with code snippets directly from the GitHub repository.
• Example outputs, Model Zoo, Citation, and License information from the GitHub README.

This model card significantly improves the discoverability and usability of the model on the Hugging Face Hub.