lime-nlp/MATH_Difficulty
收藏Hugging Face2025-04-10 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/lime-nlp/MATH_Difficulty
下载链接
链接失效反馈官方服务：
资源简介：
该数据集为MATH数学竞赛题目难度评分数据集。数据集基于Qwen 2.5-MATH-7B模型对MATH数据集中的每个问题进行难度评分，评分方法是通过模型尝试解决问题128次，计算成功次数与总尝试次数的比例，再通过公式转换为难度分数。这个评分方法可以适应性地为课程构建和模型评估提供信号。
This dataset is a difficulty scoring dataset for the MATH mathematics competition problems. It is based on the Qwen 2.5-MATH-7B models performance in attempting to solve each problem in the MATH dataset 128 times, calculating the success rate, and then converting it into a difficulty score using a formula. This scoring method provides an adaptive signal for curriculum construction and model evaluation.
提供机构：
lime-nlp
