Vision Token Manipulation Attacks on Cloud-Edge Inference of Large Vision-Language Models figure
AlphaXiv 中文概览(可滚动查看)