Precise and Dexterous Robotic Manipulation via Human-in-the-Loop Reinforcement Learning

Have not read this yet. But it’s contributed by Jeffrey Wu, who is in the Scaling Laws for Neural Language Models.