1 paper
Xuan He, Zequan Fang, Jinzhao Lian +3
The ever-increasing computation and energy demand for LLM and AI agents call for holistic and efficient optimization of LLM serving systems. In practice, heterogeneous GPU clusters…