The "maximin" function; the main player (the network) tries to "minimize" their score, assuming that the opponent (the human) tries to "maximize" their score 

This function takes the FENs parsed from Lichess and converts them into vectors to be used as training data

The model for our neural network; input layer is a vector with 768 entries to one-hot encode the board (64 squares * 6 unique pieces * 2 unique colors)