Qwen/ProcessBench

ProcessBench数据集由Qwen团队提出，旨在识别数学推理中的过程错误。该数据集包含多个分割，如gsm8k、math、olympiadbench和omnimath，每个分割对应不同的数学问题集。用户可以通过提供的代码预览数据集，并参考GitHub仓库中的评估代码和提示模板进行进一步的研究和应用。

The ProcessBench dataset, proposed by the Qwen Team, aims to identify process errors in mathematical reasoning. It includes multiple splits such as gsm8k, math, olympiadbench, and omnimath, each corresponding to different sets of mathematical problems. Users can preview the dataset using the provided code and refer to the evaluation code and prompt templates in the GitHub repository for further research and application.