TMLR-Group-HF

/

Self-Certainty-Qwen3-4B-Base-OpenRS

like 0

Follow

TMLR Group 17

This PR updates the model card for the Qwen3-4B-Base model trained on OpenRS, part of the Co-rewarding framework.

Key improvements include:

• Adding library_name: transformers to enable the automated "How to use" widget, as evidenced by the config.json and general model compatibility.
• Adding pipeline_tag: text-generation for better discoverability on the Hugging Face Hub, aligning with the model's functionality as an LLM.
• Integrating the paper link Co-rewarding: Stable Self-supervised RL for Eliciting Reasoning in Large Language Models into the model card content for proper academic attribution.

These changes enhance the model card's clarity and utility for users.