This repo gives an example of using a simple method of reinforcement learning to beat the Lunar Lander environment. The agent uses a combination of CEM and neural networks using the pytorch library.
You can quickly get this project up and running on the cloud with this button
You can checkout the accompanying Medium article here to find a step by step walkthrough on how the agent was made: https://medium.com/@donaljbyrne/landing-a-rocket-with-simple-reinforcement-learning-3a0265f8b58c