High-Dimensional Reinforcement Learning with Human Feedback