The Spelled-Out Intro to Neural Networks

Following Andrej Karpathy's Zero to Hero series — lecture 1

• 0. Intro to Marimo — Introduction to the Marimo notebook environment
• 1. Derivatives — Graphing functions and visualizing derivatives
• 2. Manual Backpropagation — Manual backprop through a scalar expression
• 3. Autograd Engine — Automated backprop, expanding tanh, verifying against PyTorch
• 4. Neural Network — Building Neuron, Layer, and MLP classes
• 5. Gradient Descent — Training loop with gradient descent
• Demo — Binary classifier on the moons dataset