ResearchBenchmarksOffline Goal-Conditioned Reinforcement Learning on puzzle 4x5 play v0Follow18Success RateARLi-0.724.14913.86May 21, 2026Evaluation ResultsMethodMethodLinksSuccess RateARLiSize=3M, Dim.=99Size=3M, Dim.=992026.0518HIQL1vrSize=3M, Dim.=99Size=3M, Dim.=992026.0517IQLSize=3M, Dim.=99Size=3M, Dim.=992026.0515ARLeSize=3M, Dim.=99Size=3M, Dim.=992026.0514HIQL2vrSize=3M, Dim.=99Size=3M, Dim.=992026.0512HIQL2vSize=3M, Dim.=99Size=3M, Dim.=992026.050