THU-KEG

/

LLaDA-8B-BGPO-math

like 1

Follow

Knowledge Engineer Group @ Tsinghua University 112

This PR improves the model card for LLaDA-8B-BGPO-math by adding important metadata:

• Adds pipeline_tag: text-generation to ensure better discoverability on the Hugging Face Hub, as this is a large language model designed for tasks like math problem solving and code generation.
• Adds library_name: transformers to enable the automated "How to use" widget on the model page. The config.json explicitly shows compatibility with the transformers library through its auto_map entries and transformers_version field, albeit likely requiring trust_remote_code=True for custom architecture loading.

No sample usage code has been added, as a straightforward, direct transformers inference snippet was not found in the associated GitHub repository's README.