TY - RPRT TI - Get More with LESS: Synthesizing Recurrence with KV Cache Compression for Efficient LLM Inference AU - Harry Dong AU - Xinyu Yang AU - Zhenyu Zhang AU - Zhangyang Wang AU - Yuejie Chi AU - Beidi Chen PY - 2024 UR - https://arxiv.org/abs/2402.09398 ID - 2402.09398 ER -