shi-labs

/

IMG

like 1

Follow

SHI Labs 87

This PR enhances the model card for "IMG: Calibrating Diffusion Models via Implicit Multimodal Guidance" by:

• Adding the pipeline_tag: text-to-image to accurately reflect the framework's goal of improving text-to-image generation alignment.
• Adding the library_name: transformers metadata, as the config.json clearly indicates the model in this repository is a LlavaLlamaForCausalLM model, compatible with the transformers library.
• Updating the paper link to the official Hugging Face Papers page: IMG: Calibrating Diffusion Models via Implicit Multimodal Guidance.
• Including a direct link to the official GitHub repository: https://github.com/SHI-Labs/IMG-Multimodal-Diffusion-Alignment.
• Incorporating the paper abstract, visual explanations (teaser and overview images), and a "Quick Start" section with code snippets for downloading checkpoints and running Gradio demos, directly extracted from the GitHub README.
• Retaining existing sections like "News", "Acknowledgements", and "Citation" for comprehensive documentation.

These updates aim to make the model more discoverable, provide essential metadata, and offer clear guidance for users on how to interact with the model.