Fetching the paper…

Fast Convergence of Natural Gradient Descent for Overparameterized Neural Networks · Around