Gen-Verse

/

MMaDA-8B-Base

like 85

Follow

Princeton-AI 112

This PR significantly enhances the model card for MMaDA-8B-Base by:

• Updating the paper link to the official Hugging Face Papers page (https://huggingface.co/papers/2505.15809).
• Adding a comprehensive "Sample Usage" section, including "Quick Start" and "Inference" examples (Text Generation, Multimodal Generation, Text-to-Image Generation) directly from the official GitHub repository, with clarifying comments for model_path.
• Including "Latest Updates" and "MMaDA Series Overview" sections from the GitHub README to provide more context about the model's development and different checkpoints.
• Adding visual elements (GIF and image) from the GitHub repository to enrich the model card.
• Adding relevant tags (e.g., diffusion-model, multimodal, text-to-image, text-generation, image-captioning, generalist-llm) and language: en to improve discoverability.

The existing library_name: transformers and pipeline_tag: any-to-any metadata are retained, as they are consistent with the model's architecture and capabilities. The license: mit is also retained based on the current model card information and majority consensus.

Please review these improvements to the model card.