ResearchBenchmarksOffline Goal-Conditioned Reinforcement Learning on HumanoidMaze Giant-Stitch v0Offline Goal-Conditioned Reinforcement Learning on HumanoidMaze Giant-Stitch v0Follow13Success RateARLi-0.522.996.510.01May 21, 2026Evaluation resultsMethodMethodLinksSuccess RateARLiSize=4M, Dim.=69Size=4M, Dim.=692026.0513ARLeSize=4M, Dim.=69Size=4M, Dim.=692026.0510HIQL1vrSize=4M, Dim.=69Size=4M, Dim.=692026.054IQLSize=4M, Dim.=69Size=4M, Dim.=692026.050HIQL2vSize=4M, Dim.=69Size=4M, Dim.=692026.050HIQL2vrSize=4M, Dim.=69Size=4M, Dim.=692026.050