MIDAS: A Dynamic Cross-GPU KV Cache Offloading Framework for LLM on GPU Cluster Systems
Zixia Zhang,Chaonong Xu,4 Authors,Chao Li