1 paper · 1 filter
Yang Yang, Qinyu Zhao, Mouxiang Chen +5
Existing scaling strategies for Multimodal Large Language Models (MLLMs) typically expand either model parameters or sequential inference computation, incurring substantial memory…