Unsupervised Curricula
         for Visual Meta-Reinforcement Learning

              Allan Jabriα Kyle Hsuβ,0 Benjamin Eysenbachγ
              Abhishek Guptaα Sergey Levineα Chelsea Finnδ
    α
      UC Berkeley β University of Toronto γ Carnegie Mellon University δ Stanford University

                         Abstract
       In principle, meta-reinforcement learning algorithms leverage experience across
       many tasks to learn fast and effective reinforcement learning (RL) strategies. How ...