Hands-On Q-Learning with Python

The standardization of RL research practice with Gym

As the field's designated standardized repository of environments, Gym models a number of classic control theory problems that we'll discuss in the next section. It also models some supervised learning problems that have structured or sequential outputs, such as in natural language processing.

In working with Gym, we'll find problems specific to robotics research, such as motion control as well as classic control theory problems and familiar arcade games. Perhaps most interestingly, we'll find environments intended to mimic algorithmic processes, such as copying characters to a tape and learning to add numbers.