ml-agents/ml-agents/mlagents/trainers/policy/policy.py

from abc import ABC, abstractmethod

from mlagents_envs.base_env import BatchedStepResult
from mlagents.trainers.action_info import ActionInfo


class Policy(ABC):
    @abstractmethod
    def get_action(
        self, batched_step_result: BatchedStepResult, worker_id: int = 0
    ) -> ActionInfo:
        pass