TorchSharedStateCritic

class maze.core.agent.torch_state_critic.TorchSharedStateCritic(networks: Mapping[Union[str, int], torch.nn.Module], num_policies: int, device: str)

One critic is shared across all sub-steps or actors (default to use for standard gym-style environments). Can be instantiated via the SharedStateCriticComposer.

property num_critics

implementation of TorchStateCritic

predict_values(observations: Dict[Union[str, int], Dict[str, torch.Tensor]]) → Tuple[Dict[Union[str, int], torch.Tensor], Dict[Union[str, int], torch.Tensor]]

implementation of StateCritic