About

A casual implementation of the Q-learning algorithm, adapted for continuous state space with Q-value function approximation using the neural network below. The heatmap on the right shows the Q-values for each player position.