The Math Underneath Neural Networks: Tensors, Matrices and PyTorch

This is the third part of a series on how language models work, built from the ground up. Part 1 built a neural network as an office building and trained it with four steps on repeat. Part 2 put text