Communication-Efficient, 2D Parallel Stochastic Gradient Descent for Distributed-Memory Optimization

Open in new window