Blockwise Adaptivity: Faster Training and Better Generalization in Deep Learning

Open in new window