/deep-Q-networks

Implementations of algorithms from the Q-learning family. Implementations inlcude: DQN, DDQN, Dueling DQN, PER+DQN, Noisy DQN, C51

Primary LanguageJupyter Notebook

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