bench-temp/mmlu-pt-undergraduate-ablation-v2-notink-natural
收藏资源简介:
该数据集是一个跨学科的问题集合,涵盖会计、建筑、工商管理、经济学、健康与生物科学、人文、语言、法律和STEM(科学、技术、工程、数学)等多个领域。数据集包含以下特征:task_raw(原始任务文本)、question_id(问题唯一标识符)、difficulty_num_correct(难度相关的正确数量)和difficulty_score(难度评分)。它可能用于教育研究、问题难度评估或模型性能测试,总大小约为1.1MB,包含多个分片,每个分片对应一个学科领域。
This dataset is a cross-disciplinary collection of questions spanning multiple fields including accounting, architecture, business administration, economics, health and biological sciences, humanities, languages, law, and STEM (science, technology, engineering, mathematics). It features columns such as task_raw (raw task text), question_id (unique question identifier), difficulty_num_correct (number correct related to difficulty), and difficulty_score (difficulty score). It is likely intended for educational research, difficulty assessment, or model performance testing, with a total size of approximately 1.1MB and multiple splits corresponding to each discipline.



