Faster Distributed Deep Net Training: Computation and Communication Decoupled Stochastic Gradient Descent

Open in new window