get_state

RLModule.get_state(components: str | Collection[str] | None = None, *, not_components: str | Collection[str] | None = None, inference_only: bool = False, **kwargs) → Dict[str, Any]

Returns the state dict of the module.

Parameters:

• components (str | Collection[str] | None) – Ignored by the default RLModule and TorchRLModule implementations. Subclasses that have sub-components, such as MultiRLModule, use this to select which sub-components to include in the returned state.

• not_components (str | Collection[str] | None) – Ignored by the default RLModule and TorchRLModule implementations. Subclasses that have sub-components, such as MultiRLModule, use this to exclude sub-components from the returned state.

• inference_only (bool) – Whether the returned state should be an inference-only state (w/o those model components that are not needed for action computations, such as a value function or a target network). Note that setting this to False might raise an error if self.inference_only is True.

• **kwargs – Forward compatible kwargs.

Returns:

This RLModule’s state dict.

Return type:

Dict[str, Any]