What does it do?

This is a reinforcement learning engine built on one of my previous games. It forces the computer to keep on playing the game and scoring the outcome every single frame. Some of the games are random, where the computer is trying to explore, or trying out the trained moves, where the computer evaluates if what it explored actually works. Random games are faster, and trained games are slower. Once the code starts it takes around 30-35 minutes to train, but can be as much as an hour. But, you can head to the trained page to see it trained fully.