bammi with better ai original game at www.bammi.com ai techinques tried deep reinforcement learning + graph convolutional network q-learning A3C policy gradient PGQL tree search minimax search monte carlo tree search with original ai as tree policy