Author name: Boyang Gong, Yu Zheng, Fanye Kong, Jie Zhou, Jiwen Lu

arXiv:2604.01989v2 Announce Type: replace
Abstract: Like a body at rest that stays at rest, we find that visual attention in multimodal large language models (MLLMs) exhibits pronounced inertia, remaining largely static once settled during early decod…

arXiv:2604.01989v1 Announce Type: new
Abstract: Like a body at rest that stays at rest, we find that visual attention in multimodal large language models (MLLMs) exhibits pronounced inertia, remaining largely static once settled during early decoding …