Example implementation of tic tac toe with tensorflow
Requires tensorflow, numpy, matplotlib
- run train.py to train the model. the pretrained model contains a model from 1000 training games
- start main to play a game against tensorflow (default is a gui app with matplotlib, but console is also possible). Switch the starting player in gui.py. Against random play the model wins ~ 97 of 100 when starting and 76 when the opponent starts (and loses 8).
The project contains the visual studio files.