Hello everyone. Mark here. This is a simple tic-tac-toe game I developed for myself as a coding challenge, written in a couple of hours, and then modified for reference.
I will add some documentation about it to make it runnable by everyone, and this will be the end of this code. Sometimes I will also update it with new versions of packages to fix some common issues and vulnerability.
Please feel free to study the code, try it yourself and engage in the issues section if you have any problems or suggestions.
To get started with this project, you would probably need the following installed on your system:
- Latest version of Node.JS
- Latest version of yarn
- Latest version of git
Also having a good rig for training neural networks is a benefit (you will get results faster).
Then you will need to clone the project:
$ git clone https://github.com/MarkusBansky/tictactoe-ml.git
$ cd tictactoe-mlThen when you are inside the folder of the project you have to install the required packages, this might take a while:
$ yarnAfter running the installation script you now should be able to start the web service locally using the built-in server:
$ yarn serveAnd the game should be available on your machine via the URL provided in the command line after running the serve action.
There also is an option to run the game from the compiled binaries. I have compiled it for Mac and Windows, both tested.
The package is available for download in the Releases section. Go to the tag for latest release and grab your binary.
The gameplay should be obvious. I will explain some features here.
The game cam save and load the model in the browser memory. It loads it automatically when you load the page when the saved model exists.
When user wins the game, this game is saved into the state, and then the model is being trained on that game. So after each time the player wins the model is being retrained to learn from the errors.
Please check out the LICENSE.md file in the repository.
Please check out the CODE_OF_CONDUCT.md file in the repository.