Toy Text#

Toy text environments are designed to be extremely simple, with small discrete state and action spaces, and hence easy to learn. As a result, they are suitable for debugging implementations of reinforcement learning algorithms.

Environments are configurable via arguments specified in each environment’s documentation.