etri-lirs/SFE

SFE（Scientists First Exam）是一个多模态数据集，旨在通过感知、理解和推理来探究多模态大语言模型（MLLM）的认知能力。数据集覆盖化学、生物、科学、地球、材料、生命和天文等多个领域，支持英文和中文两种语言。该版本是原始数据集的重新打包版本，所有图像已直接嵌入到数据文件中以便于使用和分发。

SFE (Scientists First Exam) is a multimodal dataset designed to probe the cognitive abilities of multimodal large language models (MLLM) through perception, understanding, and reasoning. The dataset covers various fields including chemistry, biology, science, earth, material, life, and astronomy, and supports both English and Chinese languages. This version is a repackaged version of the original dataset, with all images embedded directly into the data files for ease of use and distribution.