NeurIPS 2020

One Solution is Not All You Need: Few-Shot Extrapolation via Structured MaxEnt RL

Meta Review

The paper studies the interesting problem of generalization after a single task training. The idea and algorithmic development are well grounded and they lead to an algorithm that has good empirical performance against baselines. In order to improve the submission further, the authors may have a more explicit comparison of the algorithmic differences between their approach and SAC+DIAYN.