/PyDRL

A Python Deep Reinforcement Learning library. We make the whl so that you don't have to reinvent it

Primary LanguagePythonMIT LicenseMIT

PyDRL

A Python Deep Reinforcement Learning library. We make the whl so that you don't have to reinvent it

To Do

  1. Agents -
  • Random Agent
  • CEM Agent
  • DQN
  • DQN with Target Value Network
  • Double DQN
  • Dueling DQN
  • Policy Gradient
  • Actor Critic
  • Advantage Actor Critic - A2C
  • Asynchronous Advantage Actor Critic - A3C
  1. Utilities -
  • Experience Replay
  • Prioritised Experience Replay
  • Ring Buffer