遇见数据集

bench-temp/mmlu-pt-high-school-difficulty-v2-notink

收藏
Hugging Face2026-05-26 更新2026-05-31 收录
官方服务:

资源简介:

该数据集包含多个学科领域(健康与生物科学、人文、语言、STEM)的任务数据,每个数据点具有任务原始信息、问题ID、难度相关指标(如错误次数、正确次数、难度分数、难度分类)以及模型表现记录。数据集分为四个分割,分别对应不同学科,总示例数为7233,总大小约为3.18 MB。

This dataset contains task data across multiple disciplinary areas (health and biological sciences, humanities, languages, STEM), with each data point including raw task information, question ID, difficulty-related metrics (such as number of wrong and correct answers, difficulty score, difficulty bin), and model performance records. The dataset is divided into four splits corresponding to different disciplines, with a total of 7233 examples and an overall size of approximately 3.18 MB.

提供机构:
bench-temp
二维码
社区交流群
二维码
科研交流群
商业服务