Skip to main content
ASAP: Attention-Shift-Aware Pruning for Efficient LVLM Inference | Buildability Receipt | ScienceToStartup