This is a small 3 layer neural network library that I made for @MonkeyBean2 's 1000 follower contest! It uses the Numerical Gradient Descent method, which is slow, but accurate.
The AI is trained to learn how to play a match of rock-paper-scissor and win.
