AI learns to play Atari. Press D to start training. B to open info, A to close. K to open avg. fitness over generations, L to close. Z to open Network lists, X to close. Press Green flag to play against an AI.
The starting AI was a neural network that evolved/learned how to play against another hand programmed ai. Try playing against it to see how it plays. I have for the most part fixed the learning issue, using a better error function and a better mutation function fixed the issue. You still may need to restart if it does not learn after a while. Use turbowarp to speed up training.
