Foundation lessons

You know Python. These lessons make the unfamiliar objects and calls visible: a PyTorch model that changes when it learns, a neural network that can bend, how backpropagation finds every correction, and a MuJoCo world whose state changes when time advances.

The first four prepare the bouncing-ball workshop. Reinforcement learning basics starts the robot-arm workshops. Each lesson has its own small example, runnable source, and interactive figures. No video, account, server, or network is needed.