Given a checkpoint, agent needs to continue learning and take into account policy taken.
Given a checkpoint, agent needs to continue learning and take into account policy taken.