On the Generalization Gap in Reparameterizable Reinforcement Learning

OntheGeneralizationGapinReparameterizableReinforcementLearningHuanWang1StephanZheng1CaimingXiong1RichardSocher1Abstract2018a).Amodelthatperformswellinthetrainingenvi-ronment,mayormaynotperformwellw...