How to train model-based reinforcement learning agents

Create and train model-based policy optimization (MBPO) agents. An MBPO agent uses neural networks to internally approximate the environment. This reusable internal model allows for a greater sample efficiency compared to a typical model-free agent.