Scaling Down Deep Learning with MNIST-1D

Sam Greydanus, Dmitry Kobak

Although deep learning models have taken on commercial and political relevance, key aspects of their training and operation remain poorly understood. This has sparked interest in s…