Framework for deep reinforcement learning.
git clone https://github.com/ppaanngggg/DeepRLpip install -e .DoubleDQNAgent: Basic deep Q learning with double Q learning
Human-level control through deep reinforcement learning
Deep Reinforcement Learning with Double Q-learning
DDPGAgent: continue control by deep deterministic policy gradient
CONTINUOUS CONTROL WITH DEEP REINFORCEMENT LEARNING
PPOAgent: continue control by proximal policy optimization
Proximal Policy Optimization Algorithms
Replay: Basic replay, randomly choose from pool and remove the oldest one
Human-level control through deep reinforcement learning
ReservoirReplay: randomly choose from pool and randomly remove one, used in NFSPAgent's policy network
Deep Reinforcement Learning from Self-Play in Imperfect-Information Games
TmpReplay: just for module, no replay at all