Parallelization

This page contains the reference documentation for the parallelization module.

ObservationAggregator	Observation aggregator used in distributed training for aggregating observations of multiple instances of the same environment.
BaseWorker	This class holds a policy as well an env in order to step through the env, by producing action from the policy and recoding the rollout to be processed by the learner.
BaseWorkerOutput	Base class for outputs generated by the agent.

Distributed Environments

These are interfaces, classes and utility functions for distributed environments:

BaseDistributedEnv	Abstract base class for distributed environments.
DummyStructuredDistributedEnv	Creates a simple wrapper for multiple environments, calling each environment in sequence on the current Python process.

Distributed Actors

These are interfaces, classes and utility functions for distributed actors:

ActorAgent	Steps through a given environment and records rollouts.
BaseDistributedActors	The base class for all distributed actors.