Re:zero

Re:zero is a web application that can play Reversi against AI based on the AlphaZero (AZ) algorithm.

Note: Re:zero is a work in progress. In the future, Re:zero will be able to play against other players.

Features

• Play against AZ
• You can also write posts and comments on the community

System Architecture

Neural Network Architecture

Folder Structure

.
├── alphazero                   # AlphaZero
│   ├── app.py                  # Flask application (no longer used)
│   ├── dual_network.py         # Dual network that outputs the policy and value using ResNet
│   ├── evaluate_best_player.py # Evaluate the best player
│   ├── evaluate_network.py     # Evaluate the network
│   ├── game.py                 # Reversi game logic
│   ├── human_play.py           # Human play
│   ├── pv_mcts.py              # PV MCTS algorithm
│   ├── self_play.py            # Self play
│   ├── train_cycle.py          # Train cycle
│   ├── train_network.py        # Train network
│   └── model                   # Trained best model
│       └── best.h5
│
├── server                      # Laravel server
│   └── ...
├── webapp                      # Next.js web application
│   ├── public                  # Static files
│   │   └── assets              # Next.js assets
│   ├── components          # Next.js components
│   ├── lib
│   │   ├── api             # Ajax calls to the server
│   │   ├── othello         # Reversi game logic for the webapp
│   │   └── ...
│   ├── pages               # Next.js pages
│   ├── store               # Zustand store (Does not used yet)
│   ├── hooks               # Next.js hooks
│   ├── typings             # type definitions
│   └── ...
└── README.md

Built With

Frontend

• TypeScript
• Next.js
• axios
• zustand
• SWR
• Mantine
• Tailwind CSS
• TensorFlow.js

Backend

• PHP
• Laravel
• Google OAuth 2.0

AlphaZero

• Python
• TensorFlow
• Keras
• NumPy

Getting Started

This is an example of how you may give instructions on setting up Re:zero locally. To get a local copy up and running follow these simple example steps.

Prerequisites

• Node.js (version >= 14.x)
• PHP (version >= 8.1)
• Python (version >= 3.9)
• Yarn (version >= 1.22) (recommended)

Installation

1. Clone the repository

git clone https://github.com/1ncursio/laravel-othello.git

2. Install NPM dependencies for the frontend

cd laravel-othello/webapp
yarn install

3. Run the frontend server

yarn dev

4. Install Composer dependencies for the backend

cd laravel-othello/server
composer require

5. Migrate the database

php artisan migrate

6. Copy .env.example to .env and fill in the required values

cp .env.example .env

7. Run the backend server

php artisan serve

Then open your browser and navigate to http://localhost:3000/

Future Plans

• Refactor the code to use the Zustand
• Write tests
• Deploy the project
• Implement multi-player mode
• Migrate Backend Framework (e.g. Nest.js, fastify, django, etc.)
• Support i18n
• Support PWA
• Implement Scalable Game Board (e.g. 6x6, 8x8, 10x10, etc.)

License

Distributed under the MIT License. See LICENSE for more information.

Contact