1 paper · 1 filter
Luozheng Qin, Jia Gong, Qian Qiao +6
Unified multimodal models integrating visual understanding and generation face a fundamental challenge: visual generation incurs substantially higher computational costs than under…