UNO Card Game with Reinforcement Learning๏
Welcome to the documentation for the UNO Card Game RL project! This project implements a complete UNO card game with multiple Reinforcement Learning agents.
Overview๏
This project explores the application of Reinforcement Learning to the classic UNO card game. Our agents achieve 60%+ win rates against random opponents, with our Self-Play Champion targeting 70%+ win rates.
Key Features๏
๐ Complete UNO Engine: Full game rules including special cards
๐ค Multiple RL Agents: Q-Learning, DQN, PPO, Recurrent PPO
๐ฎ Modern GUIs: Pygame-based interfaces with glassmorphism design
๐ฅ Multiplayer Support: 2-4 player battles
๐ Model Comparison: Evaluate and compare different agents
๐ Self-Play Training: Train agents against themselves
Quick Start๏
# Clone repository
git clone https://github.com/user/uno-card-game-rl.git
cd uno-card-game-rl
# Install dependencies
pip install -r requirements.txt
# Play against AI
python uno_gui.py
# Watch AI battles
python model_battle_gui.py
Table of Contents๏
Getting Started
User Guide
Training
Developer Guide