Reinforcement Learning Minesweeper

Reinforcement learning program playing Minesweeper, created using StableBaselines3, Gymnasium, CUDA and Pytorch. Documented and compared algorithms: MaskablePPO, PPO and A2C.

Requirements

• Windows
• Python 3.11
• Miniconda
• CUDA 12.4 (Optional)

Miniconda setup

• During or after installation set enviornement variable to .../miniconda3 directory
• Create enviornement conda create .../python-minesweeper-main/.conda or Create enviornement... option in Visual Studio Code
• Activate enviornement activate
• pip install -r requirements.txt