rzzhan

/

ExGRPO-Qwen2.5-Math-1.5B-Zero

like 0

This PR significantly enhances the model card for the ExGRPO-Qwen2.5-Math-1.5B-Zero model.

Key improvements include:

• Addition of pipeline_tag: text-generation for better discoverability and Hub integration.
• Addition of library_name: transformers to enable the automated "How to use" widget, reflecting the model's compatibility and facilitating usage.
• Linking the model to its paper: ExGRPO: Learning to Reason from Experience.
• Including links to the official GitHub repository and the Hugging Face Collection.
• Incorporating the paper's abstract and key highlights to provide an overview of the ExGRPO framework.
• Adding a Python code snippet for sample usage with the transformers library, specifically tailored to the model's chat template for reasoning tasks, along with an example of the expected output structure.
• Embedding key evaluation result images from the GitHub repository for a quick performance overview.

Please review and merge if these updates align with the repository's goals.