Tweeted By @svlevine

Meta-learning enables fast learning, but needs hand-engineered meta-training tasks. Can we get the tasks themselves automatically? Our first attempt at this for RL: unsupervised meta-reinforcement learning:https://t.co/ePkIsMcMjZ
w/ Abhishek Gupta, Ben Eysebach, @chelseabfinn

— Sergey Levine (@svlevine) June 13, 2018