ResearchBenchmarksReinforcement Learning on Atari Beam RiderFollow188,200Mean ReturnR2D210,04856,299102,550148,801Feb 14, 2020Evaluation ResultsMethodMethodLinksMean ReturnR2D22020.02188,200NGUN=1, Variant=without RNDN=1, Variant=without RND2020.02166,500NGUN=1N=12020.02114,600R2D2Loss=RetraceLoss=Retrace2020.02111,100R2D2Extension=RNDExtension=RND2020.0285,700NGUN=32N=322020.0268,700Human2020.0216,900