Neural Network that learns to play La Vieja (Tic Tac Toe) from Zero

The idea is to simulate a neural network to play tic tac toe against itself, at first it will know nothing about the game, but it will store every step of the game, and check the winner and losser set of moves, and infere from those good and bad strategies.

How to use

alvieja --name "alvieja" train alvieja --name "alvieja" test