Kwai-Klear/qwen2.5-math-rlep · Enhance model card with metadata, paper link, and usage example

Jul 18

This PR significantly improves the model card for Kwai-Klear/qwen2.5-math-rlep by adding comprehensive information and crucial metadata.

Specifically, it:

Updates the paper link to the official Hugging Face Papers page: RLEP: Reinforcement Learning with Experience Replay for LLM Reasoning.
Adds key metadata: pipeline_tag: text-generation, library_name: transformers, license: apache-2.0, and specifies datasets and base_model for better context and discoverability.
Includes a direct link to the official GitHub repository for code and further details.
Provides an overview of the RLEP framework and its key highlights from the paper's abstract and GitHub README.
Showcases the model's superior performance with an evaluation table and illustrative images.
Adds a practical Python code example for quick inference using the EaModel as described by the project.

This update will greatly enhance the usability and discoverability of the model on the Hugging Face Hub.

Enhance model card with metadata, paper link, and usage example7b7070dc

hongzhizhang

Klear team, Kuaishou Technology org Jul 18

Hi Nielsr, thank you for the pull request which make the readme better. I will validate the ea_model inference script and merge the PR today. I evalute the model with vllm, I would also include the corresponding evaluate demo code soon.

hongzhizhang changed pull request status to merged Jul 18

hongzhizhang

Klear team, Kuaishou Technology org Jul 18

I’ve added the vLLM‑based inference snippet to the model card and removed the earlier EaModel demo for simplicity. If you think keeping the EaModel example would benefit the community, I’m happy to include it as well.