TMLR-Group-HF

/

Self-Certainty-Qwen3-1.7B-Base-MATH

like 1

Follow

TMLR Group 20

This PR enhances the model card for TMLR-Group-HF/Self-Certainty-Qwen3-1.7B-Base by:

• Adding pipeline_tag: text-generation for improved discoverability on the Hub, as it's a causal language model.
• Adding library_name: transformers to enable the automated "How to use" widget, based on its Qwen3ForCausalLM architecture and transformers_version in config.json.
• Including relevant tags such as qwen3, fine-tuning, reinforcement-learning, reasoning, mathematical-reasoning, and self-supervised-learning.
• Integrating a direct link to the paper: Co-rewarding: Stable Self-supervised RL for Eliciting Reasoning in Large Language Models.
• Correcting the GitHub repository link to https://github.com/tmlr-group/Co-rewarding, which was previously outdated.
• Clarifying the model's description, noting its role as a baseline model trained with the Self-Certainty method within the context of the Co-rewarding paper.
• No sample usage is included, as no direct inference snippet was found in the provided GitHub README content, adhering to the specified guidelines.