Abstract: Model Free Reinforcement Learning (MFRL) has shown significant promise for learning dexterous robotic manipulation tasks, at least in simulation. However, the high number of samples, as well ...