[PYTORCH] Deep Q-learning for playing Tetris

Introduction

Here is my python source code for training an agent to play Tetris. It could be seen as a very basic example of Reinforcement Learning's application.

Tetris demo

The demo could also be found at youtube demo

How to use my code

With my code, you can:

• Train your model from scratch by running python train.py
• Test your trained model by running python test.py

Trained models

You could find my trained model at trained_models/tetris

Requirements

• python 3.6
• PIL
• cv2
• pytorch
• numpy
• matplotlib