tf_agents.bandits.policies.constraints.InputNetworkConstraint

Class for representing a constraint using an input network.

Inherits From: BaseConstraint

tf_agents.bandits.policies.constraints.InputNetworkConstraint(
    time_step_spec: tf_agents.typing.types.TimeStep,
    action_spec: tf_agents.typing.types.BoundedTensorSpec,
    input_network: Optional[tf_agents.typing.types.Network] = None,
    name: Optional[Text] = 'InputNetworkConstraint'
)

This constraint class uses an input network to compute the action feasibility. It assumes that the input network is already trained and it can be provided at construction time or later using the set_network() function.

Args
time_step_spec	A TimeStep spec of the expected time_steps.
action_spec	A nest of BoundedTensorSpec representing the actions.
input_network	An instance of tf_agents.network.Network used to provide estimates of action feasibility.
name	Python str name of this agent. All variables in this module will fall under that name. Defaults to the class name.

Attributes
network

Methods

compute_loss

View source

compute_loss(
    observations: tf_agents.typing.types.NestedTensor,
    actions: tf_agents.typing.types.NestedTensor,
    rewards: tf_agents.typing.types.Tensor,
    weights: Optional[types.TensorOrArray] = None,
    training: bool = False
) -> tf_agents.typing.types.Tensor

__call__

View source

@abc.abstractmethod
__call__(
    observation, actions=None
)

Returns the probability of input actions being feasible.