Entropy-regularised Reinforcement Learning