AlphaZero: A DeepMind Algorithm Replication

Project Title and Description

This project is aimed at replicating DeepMind's AlphaZero algorithm, specifically for the game of Connect 4. The project uses a combination of Monte Carlo Tree Search (MCTS) and a deep convolutional neural network to train an AI agent capable of playing Connect 4 at a high level.

Installation and Setup

1. Clone the repository: git clone https://github.com/Honyant/AlphaZero.git
2. Navigate to the project directory: cd AlphaZero
3. Install the required Python packages: pip install -r requirements.txt

Usage and Examples

To train the AI, run the train.py file:

python train.py

To train the AI in parallel, run the parallel_train.py file:

python parallel_train.py

To play against the AI, run the ai_game.py file:

python ai_game.py

Acknowledgements and Credits

This project was inspired by DeepMind's AlphaZero algorithm. It uses the PyTorch library for the neural network implementation.