This repository contains a Actor-Critic model as well as a genetic algorithm to master the game of pong. It also contains a hypernetwork to search for low loss-valleys on the losssurface, but there still needs to be a lot of work done.
This again needs some work done, but for now, the Actor-Critic-Model works well as well as the genetic approach. I will upload models of the corresponding agents soon.