Can Adam optimizer be used with torch.multiprocessing?

1

Trying to train a network using multiple CPUs but I am not sure if Adam optimizer can be used instead of SGD, should the optimizer be shared??