THU-KEG

/

LLaDA-8B-BGPO-countdown

like 1

Follow

Knowledge Engineer Group @ Tsinghua University 103

This PR enhances the model card for the LLaDA-8B-BGPO-countdown model by:

• Adding pipeline_tag: text-generation to improve discoverability on the Hugging Face Hub, as the model is a diffusion large language model (dLLM) designed for tasks like planning, math, and code generation.
• Adding library_name: transformers based on the config.json (transformers_version and auto_map entries) and the convert_to_hf.sh script mentioned in the GitHub repository, enabling automated usage snippets on the Hub.
• Enriching the markdown content with detailed sections from the GitHub README, including "Performance" (with direct links to result images), "Acknowledgments", and "Citation" for a more complete and informative model card.

Please review and merge this PR.