Tags: #machine-learning

Softmax to MNIST: Building a Tiny Autograd and Classifier

Deriving softmax, log-softmax, and NLL loss, then training an MLP on MNIST with a home-grown autograd.