Fast reinforcement learning through the composition of behaviours Deep Mind, 12 Oct 2020 Imagine if you had to learn how to chop, peel and stir all over again every time you wanted to learn a new recipe. In many…