TIGER-Lab/rStar-Critique-Data

rStar-Critique-Data数据集是用于增强编码模型的一种数据集，它遵循Critique Reinforcement Learning（CRL）范式，通过训练模型生成对（问题，解决方案）对的批评来提高编码模型的性能。该数据集是论文《Critique-Coder: Enhancing Coder Models by Critique Reinforcement Learning》中研究成果的一部分。

The rStar-Critique-Data dataset is designed to enhance coding models, adhering to the Critique Reinforcement Learning (CRL) paradigm. It improves coding model performance by training the model to generate critiques for (question, solution) pairs. This dataset is part of the research presented in the paper Critique-Coder: Enhancing Coder Models by Critique Reinforcement Learning.