1 paper
Foteini Strati, Zhendong Zhang, George Manos +7
The high GPU demand of ML training makes it hard to allocate large homogeneous clusters of high-end GPUs in a single availability zone. Leveraging heterogeneous GPUs available with…