FlowCut: Rethinking Redundancy via Information Flow for Efficient Vision-Language Models

Open in new window