Environments is a series of works focused on the visualization of neural network training through reinforcement learning techniques.Reinforcement learning is one of the three most important paradigms of machine learning together with supervised learning and unsupervised learning. This technique allows artificial intelligence to learn how to perform actions that lead to achieving a goal by acting in the most effective way possible within a given environment.In Environments each work shows a small world within which the intelligent agent can operate without constraints. The AI has no prior knowledge of the rules and of what surrounds it, but after various attempts to get the right feedback from the environment will learn by itself to move as it best suits it to get the reward.Reinforcement learning is often modeled through Markov decision processes (MDPs), which are particularly useful for addressing a wide range of optimization problems in contexts where results are partly random and partly under the control of a decision maker.