deep-Q-networks

Implementation of algorithms from the Q-learning family

Check /ipynb for jupiter notebooks of implementations

Implemented:

• Vanilla DQN
• Double DQN
• Dueling DQN
• Prioritized experience replay + DDQN
• Noisy DQN (in progress)
• C51 Categorical DQN (in progress)
• DRQN with Gated Recurrent Units (in progress)

Dependencies:

• PyTorch = 0.4.1
• Python = 3.6
• Gym = 0.12.5

Associated Papers:

To be added