MS-Bench

This is MS-Bench, the first comprehensive benchmark co-developed with archaeologists, comprising 5,076 high-resolution images from 4th to 14th century and 9,982 expert-curated questions across nine sub-tasks aligned with archaeological workflows. Through four prompting strategies, we systematically evaluate 32 LMMs on their: effectiveness robustness cultural contextualization

本数据集为MS-Bench，是首个与考古学家联合研发的综合性基准测试集，包含5076张公元4世纪至14世纪的高分辨率图像，以及9982道经专家精心编撰的题目，涵盖9个契合考古工作流程的子任务。本研究通过四种提示策略，系统评估了32个多模态大模型（Large Multimodal Model，LMM）的三项核心性能：有效性、鲁棒性与文化语境适配能力。