Utilities

Replay Buffer

The replay_buffer.py provides 2 sorts of replay buffers as follows.

class TransReplayBuffer(size) is the replay buffer that stores the transition snippet such as $(s_{t}, a_{t}, r_{t}, s_{t+1})$. size is for the replay buffer size.

class EpisodeReplayBuffer(size) is the replay buffer that stores the complete episode such as $(s_{0}, a_{0}, r_{0}, s_{1}, a_{1}, r_{1}, s_{2}, ..., s_{T})$. size is for the replay buffer size.

Tester

The tester.py provides class PGTester(args, behaviour_net, env, render=False) that is a driver for testing the trained algorithms. If you would know how to use this class, please see test.py for details.

Trainer

The trainer.py provides class PGTrainer(args, model, env, logger) that is a driver for training algorithms. If you would know how to use this class, please see train.py for details.