This demo demonstrates training an AI model to beat the iconic 1990s arcade game Double Dragon. The video delves into the principles of reinforcement learning and showcases advanced algorithms such as Proximal Policy Optimization (PPO) in action.
The presentation covers the mechanics of the training process, including the reward systems that drive the AI’s behavior. It provides a detailed look at how these algorithms are trained within OpenShift AI, utilizing powerful open-source tools.
Moreover, the video offers a practical guide on setting up the OpenShift AI environment with a custom image to incorporate open-source tools such as PyBoy and Gym. This section provides step-by-step instructions, making it easy for you to configure your environment for optimal performance and integration.
Checkout the PokeRL community that has used the same approach to beat Pokemon Red
https://github.com/PWhiddy/PokemonRedExperiments
