Video-MMMU

A multi-discipline benchmark evaluating large multimodal models' ability to acquire and apply knowledge from expert-level professional videos across six disciplines through three cognitive stages (Perception, Comprehension, Adaptation), measured by question-answering accuracy.