Solving hard-exploration environments in an important challenge in reinforcement learning. Several approaches have been proposed and studied, such as intrinsic motivation, co-evolution of agents and tasks, and mu