This contains the source code for the Hexapawn example of Chapter 4.

Execute the files as follows:

- create a command prompt and change to the this directory:

# generate training data for supervised learning

# by solving the game with minimax

python mnx_generateTrainingData.py

# train the network with supervised learning from the data

# that was created using minimax search

python sup_network.py

# evalute the supervised network

python sup_eval.py

# train the network using reinforcement learning

# using MCTS, i.e. in "AlphaZero-style"

python rnf_train.py

# evaluate the network that was trained

# by reinforcement learning

python rnf_eval.py

from common.game import Board

from mnx_minimax import minimax