David MortonLearning Disentangled Representations — Part 1 (simple dots)After reading the fantastic post, What a Disentangled Net We Weave, I wanted to try it out for myself. The post makes no mention of code…Jun 14, 20181Jun 14, 20181
David MortonIncreasing Mini-batch Size without Increasing MemoryThe most popular technique used to train neural networks today is gradient descent, where the error of the network is minimized by…Apr 9, 20182Apr 9, 20182
David MortonExploring the Learning Rate SurveyThis post will showcase a method to gather important information about how a deep learning model behaves under different learning rates…Apr 3, 2018Apr 3, 2018