DefaultPolicy¶

class maze.core.agent.default_policy.DefaultPolicy(policies: Union[List[Union[None, Mapping[str, Any], Any]], Mapping[Union[str, Type], Union[None, Mapping[str, Any], Any]]])¶

Encapsulates one or more policies identified by policy IDs.

Parameters: policies – Dict of policy IDs and corresponding policies.

compute_action(observation: Dict[str, numpy.ndarray], maze_state: Optional[Any] = None, env: Optional[maze.core.env.base_env.BaseEnv] = None, actor_id: Optional[maze.core.env.structured_env.ActorID] = None, deterministic: bool = False) → Dict[str, Union[int, numpy.ndarray]]¶

(overrides Policy)

implementation of Policy interface

compute_top_action_candidates(observation: Dict[str, numpy.ndarray], num_candidates: Optional[int], maze_state: Optional[Any], env: Optional[maze.core.env.base_env.BaseEnv], actor_id: Optional[maze.core.env.structured_env.ActorID] = None) → Tuple[Sequence[Dict[str, Union[int, numpy.ndarray]]], Sequence[float]]¶

(overrides Policy)

implementation of Policy interface

needs_state() → bool ¶

(overrides Policy)

This policy does not require the state() object to compute the action.

policy_for(actor_id: Optional[maze.core.env.structured_env.ActorID]) → maze.core.agent.flat_policy.FlatPolicy ¶

Return policy corresponding to the given actor ID (or the single available policy if no actor ID is provided)

Parameters: actor_id – Actor ID to get policy for
Returns: Flat policy corresponding to the actor ID

seed(seed: int) → None ¶

(overrides Policy)

Not applicable since Global seed should already be set before initializing the models