1 paper
Chengtao Lv, Jinyang Du, Shuyi Feng +7
World Action Models (WAMs) incorporate visual representations from video generation backbones to guide action prediction. Recent efficient WAMs adopt Mixture-of-Transformers (MoT)…