Fetching the paper…

Layer-Condensed KV Cache for Efficient Inference of Large Language Models · Around