Partitioning sparse deep neural networks for scalable training and inference | Read Paper on Bytez