Efficient Video Sampling: Pruning Temporally Redundant Tokens for Faster VLM Inference Paper • 2510.14624 • Published Oct 16 • 1