Fetching the paper…

SparseVLM: Visual Token Sparsification for Efficient Vision-Language Model Inference · Around