reward model ensemble

retraining (1) interaction-efficient (1) target environment (1) conservative optimization (1) evolution strategies (1) best-of-n sampling (1) proximal policy optimization (1) reinforcement learning from human feedback (1) imitation learning (1)