1 paper · 1 filter
Jinghao Wang, Yihang Zhou, Xiaoyang Sun +5
Modern GPU clusters must simultaneously serve deep learning training and offline large language model inference workloads, yet existing schedulers treat these as isolated resource…