This is a my practical project for A-Level Computer Science. It is a simple Python implementation of AlphaZero, one of the strongest chess-playing models to date. It utilises reinforcement learning to train itself, eliminating the need for expensive data and human resources. For more information, please see DeepMind's blog.
- Single player against AI
- Two player game
- Save and replay games
- Self-play demonstration
- User Interface customisation
For video demo please check out my testing playlist on YouTube
 |
 |
|---|---|
 |
 |
In terminal, run the following command:
pip install pygame torch numpy
Before running main.py, you need to run train.py to create a .pth file. This file contains the weight used by the neural network.
Run main.py to enter the main user interface, or run train.py to train the model.