A Provably Efficient Model-Free Posterior Sampling Method for Episodic Reinforcement Learning | Read Paper on Bytez