xiaomi-research

/

lego-edit

like 1

Follow

Xiaomi Research 11

This PR significantly enhances the model card for Lego-Edit by:

• Updating metadata:
  • Changing the license to cc-by-nc-4.0 to accurately reflect the license for the model weights, as specified in the GitHub repository's disclaimer.
  • Adding the pipeline_tag: image-to-image to improve model discoverability, aligning with its core functionality as an image editing framework.
  • Adding library_name: transformers based on the model's Qwen2_5_VLForConditionalGeneration architecture and related processor/tokenizer classes found in the config.json and tokenizer_config.json files, enabling automated usage snippets on the Hub.
  • Adding relevant tags like vision-language, image-editing, multimodal, and qwen2_5 for better categorization.
• Adding comprehensive content:
  • Including the model's abstract, features, and detailed usage instructions (Quick Start, New Tools Integration, More Usages) directly from the official GitHub README.
  • Providing explicit links to the paper (Lego-Edit: A General Image Editing Framework with Model-Level Bricks and MLLM Builder), the project page (https://xiaomi-research.github.io/lego-edit/), and the GitHub repository (https://github.com/xiaomi-research/lego-edit).
  • Incorporating the original GitHub repository's visual elements, disclaimer, citation information, and acknowledgments to provide a complete overview.

This update makes the model card much more informative and accessible for users on the Hugging Face Hub.