ResearchBenchmarksOffline Goal-Conditioned Reinforcement Learning on cube-triple-play v0Follow15Average Binary Success RateARLe-0.63.457.511.55May 21, 2026Evaluation ResultsMethodMethodLinksAverage Binary Success RateARLeSize=3M, Dim.=46Size=3M, Dim.=462026.0515ARLiSize=3M, Dim.=46Size=3M, Dim.=462026.0514IQLSize=3M, Dim.=46Size=3M, Dim.=462026.0511HIQL1vrSize=3M, Dim.=46Size=3M, Dim.=462026.057HIQL2vrSize=3M, Dim.=46Size=3M, Dim.=462026.051HIQL2vSize=3M, Dim.=46Size=3M, Dim.=462026.050