Simple DQN example using Pytorch (Covert tensorflow version from this Github Page).

• Natural Deep Q learning
• Deep Q learning with Prioritized Experience Replay
• Double Deep Q learning
• Dueling Deep Q learning

The above implementations are integrated together, which can be selected in the initialization of DQN network.

• Python 2.7
• Pytorch 0.4.1

• Download the three .py files
• Run 'run_this.py' and see the result of Deep Reinforcement learning in small game.

Morvan Zhou - Reinforement learning tutorials