jiedabawang/mmlu

该数据集名为Measuring Massive Multitask Language Understanding (MMLU)，是一个用于测量大规模多任务语言理解的数据集。它包含多种学科的多个选择题问题，涵盖从抽象代数到世界历史的广泛主题。数据集设计用于问答任务，特别是多项选择题回答。每个主题作为一个独立的配置，包含问题、主题、选项和答案等特征。数据集为单语（英语），并包含测试集、验证集和开发集。

The dataset named Measuring Massive Multitask Language Understanding (MMLU) is designed for measuring massive multitask language understanding. It includes multiple-choice questions across a wide range of subjects, from abstract algebra to world history. The dataset is intended for question-answering tasks, specifically multiple-choice QA. Each subject is a separate configuration with features such as question, subject, choices, and answer. The dataset is monolingual (English) and includes splits for test, validation, and development sets.