Sparse Attention Remapping with Clustering for Efficient LLM Decoding on PIM

Re-assign community