Berkeley Deep Reinforcement Learning cs294 solution - View it on GitHub