AIBench/EESE

Ever-Evolving Science Exam (EESE)是一个动态的科学能力评估基准，旨在可靠地评估基础模型中的科学能力。EESE-V2是其更新版本，改进了问题的质量和范围以及问题的差异度。数据集支持封闭式（单选/多选、填空、判断对错）和开放式的问题回答任务，并允许跨不同学术学科和子领域进行评估。数据集支持英语，包含486个问题的EESE版本和500个问题的EESE-V2版本。数据集的学科分布涵盖农业科学、自然科学、工程技术科学、医学科学以及人文社会科学。

The Ever-Evolving Science Exam (EESE) is a dynamic benchmark designed to reliably assess scientific capabilities in foundation models. EESE-V2, the updated version, improves the quality and range of questions and the level of difficulty. The dataset supports both closed-ended (single/multiple choice, fill-in-the-blank, true-or-false) and open-ended question answering tasks, and allows for assessment across different academic disciplines and subfields. It supports English language and consists of the EESE version with 486 questions and the EESE-V2 version with 500 questions. The subject distribution of the dataset covers Agricultural Sciences, Natural Sciences, Engineering and Technological Sciences, Medical Sciences, and Humanities and Social Sciences.