Sutton & Barto · Reinforcement learning: an introduction

Learn by trial and error

Sixteen algorithms from the book, from bandits to actor–critic, each running live in your browser. Change the step size, exploration or discount and watch what the agent learns.