Get More with LESS: Synthesizing Recurrence with KV Cache Compression for Efficient LLM Inference

Re-assign community