viarect
dennybritz/reinforcement-learning