R-Sparse: Rank-Aware Activation Sparsity for Efficient LLM Inference

Re-assign community