DeltaStateCriticComposer

class maze.perception.models.critics.delta_state_critic_composer.DeltaStateCriticComposer(observation_spaces_dict: Dict[str | int, gymnasium.spaces.Dict], agent_counts_dict: Dict[str | int, int], networks: List[None | Mapping[str, Any] | Any] | Mapping[str | Type, None | Mapping[str, Any] | Any])

First sub step gets a regular critic, subsequent sub-steps predict a delta w.r.t. to the previous critic.

Instantiates a TorchDeltaStateCritic.

Parameters:
  • observation_spaces_dict – Dict of sub-step id to observation space.

  • networks – The single, shared critic network as defined in the config.

property critic: TorchDeltaStateCritic

(overrides BaseStateCriticComposer)

implementation of BaseStateCriticComposer