Open field notes for reinforcement learning

Understand the world before you train the agent.

Practical, step-by-step environment specifications—observations, actions, rewards, edge cases and baselines—written to be built from.