ray.rllib.env.env_runner_group.EnvRunnerGroup#

class ray.rllib.env.env_runner_group.EnvRunnerGroup(*, env_creator: Callable[[EnvContext], Any | gymnasium.Env | None] | None = None, validate_env: Callable[[Any | gymnasium.Env], None] | None = None, default_policy_class: Type[Policy] | None = None, config: AlgorithmConfig | None = None, num_env_runners: int = 0, local_env_runner: bool = True, logdir: str | None = None, _setup: bool = True, tune_trial_id: str | None = None, num_workers=-1, local_worker=-1)[source]#

Set of EnvRunners with n @ray.remote workers and zero or one local worker.

Where: n >= 0.

DeveloperAPI: This API may change across minor Ray releases.

Methods

`__init__`	Initializes a EnvRunnerGroup instance.
`add_policy`	Adds a policy to this EnvRunnerGroup's workers or a specific list of workers.
`add_workers`	Creates and adds a number of remote workers to this worker set.
`fetch_ready_async_reqs`	Get esults from outstanding asynchronous requests that are ready.
`foreach_env`	Calls `func` with all workers' sub-environments as args.
`foreach_env_with_context`	Calls `func` with all workers' sub-environments and env_ctx as args.
`foreach_policy`	Calls `func` with each worker's (policy, PolicyID) tuple.
`foreach_policy_to_train`	Apply `func` to all workers' Policies iff in `policies_to_train`.
`foreach_worker`	Calls the given function with each EnvRunner as its argument.
`foreach_worker_async`	Calls the given function asynchronously with each worker as the argument.
`foreach_worker_with_id`	Calls the given function with each EnvRunner and its ID as its arguments.
`get_spaces`	Infer observation and action spaces from one (local or remote) EnvRunner.
`healthy_worker_ids`	Returns the list of remote worker IDs.
`is_policy_to_train`	Whether given PolicyID (optionally inside some batch) is trainable.
`num_healthy_remote_workers`	Returns the number of healthy remote workers.
`num_healthy_workers`	Returns the number of all healthy workers, including the local worker.
`num_in_flight_async_reqs`	Returns the number of in-flight async requests.
`num_remote_env_runners`	Returns the number of remote EnvRunners.
`num_remote_worker_restarts`	Total number of times managed remote workers have been restarted.
`num_remote_workers`	Returns the number of remote EnvRunners.
`probe_unhealthy_workers`	Checks for unhealthy workers and tries restoring their states.
`reset`	Hard overrides the remote EnvRunners in this set with the provided ones.
`stop`	Calls `stop` on all rollout workers (including the local one).
`sync_env_runner_states`	Synchronizes the connectors of this EnvRunnerGroup's EnvRunners.
`sync_weights`	Syncs model weights from the given weight source to all remote workers.

Attributes

local_env_runner

Returns the local EnvRunner.