Implementation of the AlphaZero algorithm for qubic - (3 Dimensional Tic Tac Toe) and connect4.

Authors: Sasidharan Mahalingam (Overall framework integration, modyfing and fixes bugs in the alpha zero implementation) Rafael Espericueta (Implementation of the Simple Heuristic Agent) Eliana Stefani (Implementation of MiniMax Agent)

Packages Required: Python - 3.5 or later Tensorflow - 1.12 or later (Tensorflow-gpu with a working GPU accelarator recommended) Atleast 400 GB of free space on disk recommeded to saved the models trained

Instructions to Train an AlphaZero Agent for Connect4: python main.py --game connect4

Instructions to Train an AlphaZero Agent for Qubic: python main.py --game qubic

Instructions to run the trained model for connect4: python pit.py -p -o (random, heuristic, minimax, alphazero, human)

Instructions to run the trained model for qubic: python pit_qubic.py -p -o (random, heuristic, minimax, alphazero, human)

Acknowledgement:
The framework is based on the implementation of Surag Nair for the game Othello