bench-temp/mmlu-pt-undergraduate-ablation-v2-notink-equal

该数据集是一个多学科领域的问答数据集，涵盖会计、建筑、商业管理、经济学、健康与生物科学、人文、语言、法律和STEM（科学、技术、工程和数学）等9个学科分片。每个数据样本包含任务原始文本（task_raw）、问题ID（question_id）、难度正确数（difficulty_num_correct）和难度得分（difficulty_score）等特征。数据集总大小约为632KB，包含超过10,000个样本，旨在支持跨学科的问题难度评估或相关NLP任务。This dataset is a multi-disciplinary question-answering dataset covering nine subject splits: accounting, architecture, business administration, economics, health and biological sciences, humanities, languages, law, and STEM (science, technology, engineering, and mathematics). Each data sample includes features such as task raw text (task_raw), question ID (question_id), difficulty number correct (difficulty_num_correct), and difficulty score (difficulty_score). The total dataset size is approximately 632KB with over 10,000 examples, designed to support cross-disciplinary difficulty assessment or related NLP tasks.