albertqjiang / alphagozero-othello

Play the game of Othello with methods used by AlphaGo Zero

alphagozero-othello

Play the game of Othello with methods used by AlphaGo Zero

Prerequisite:

Standard Machine Learning Python libraries MXNet/Gluon: installation here: installation

How to use:

First execute mcts.py, change the code

state = OthelloState(16)

in line 383, default is 16 * 16. This will run the game of Othello with MCTS and record the states/probabilities. Change the code at the bottom to run it for multiple times.

Then execute the jupyter notebook mcts_result_formulation.ipynb to handle the records. By default this will only handle the result for the first game.

Execute ResNet_simplified.ipynb to run the simplified resnet on the results for 1000 iterations.