The Impact of Neural Network Overparameterization on Gradient Confusion and Stochastic Gradient Descent

Open in new window