ResearchBenchmarksOffline Goal-Conditioned Reinforcement Learning on humanoidmaze giant-navigate v0Offline Goal-Conditioned Reinforcement Learning on humanoidmaze giant-navigate v0Follow81Success RateHSVL-3.2418.6340.562.37Apr 19, 2026Apr 24, 2026Apr 29, 2026May 5, 2026May 10, 2026May 15, 2026May 21, 2026Evaluation resultsMethodMethodLinksSuccess RateHSVL2026.0481ARLiSize=4M, Dim.=69Size=4M, Dim.=692026.0549ARLeSize=4M, Dim.=69Size=4M, Dim.=692026.0538HIQL1vrSize=4M, Dim.=69Size=4M, Dim.=692026.0522HIQL2026.0412HIQL2vSize=4M, Dim.=69Size=4M, Dim.=692026.0512HIQL2vrSize=4M, Dim.=69Size=4M, Dim.=692026.0511CRL2026.043QRL2026.041IQLSize=4M, Dim.=69Size=4M, Dim.=692026.051GCBC2026.040GCIVL2026.040GCIQL2026.040