Galaxian Game Reinforcement Learning Project

Overview

This project focuses on training an agent to play the Galaxian game using Deep Reinforcement Learning (DRL). The agent employs advanced neural network architectures and techniques to enhance its decision-making capabilities in the dynamic game environment.

Installation

1. Clone the repository:

   https://github.com/khashayarghamati/Galaxian-Game-Reinforcement-Learning.git

2. Create and activate a Conda environment:

   conda env create -f environment.yml
   conda activate gal3

3. Set up Neptune API Token and Project Name:

   • Add your Neptune API token and project name to the .env_temp file:

     NEPTUNE_TOKEN=your_api_token
     NEPTUNE_PROJECT=your_project_name

   • Rename the file from .env_temp to .env.

Usage

1. Run the training script:

   python main.py

2. Once training is complete, run the agent in the game environment:

   • Edit replay.py and pass your checkpoint file path to Path() in that file.

      # Edit this line in replay.py
      path = Path('your_checkpoint_path.chkpt')

   • Run replay.py:

     python replay.py

   • After running, it will generate a vid.mp4 video for you to observe the training progress.