CASP: Compression of Large Multimodal Models Based on Attention Sparsity

Re-assign community