keskarnitish

Code to reproduce some of the figures in the paper "On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima"

Optimization using Stochastic quasi-Newton methods