1 paper
Shengjie Jin, Zelong Sun, Hengbo Xu +2
Vision-Language Models (VLMs) enable autonomous GUI navigation, but agents still struggle to process and learn from dense, continuous visual histories. This bottleneck hinders both…