1 paper
Anxhelo Xhebraj, Sean Lee, Hanfeng Chen +1
We present JaxPP, a system for efficiently scaling the training of large deep learning models with flexible pipeline parallelism. We introduce a seamless programming model that all…