ResearchBenchmarksOffline Goal-conditioned Reinforcement Learning on humanoidmaze medium-navigate-oraclerep v0Follow91Task 1 ScoreCRL0.5224.0147.570.99Oct 26, 2025Evaluation ResultsMethodMethodLinksTask 1 ScoreTask 2 ScoreTask 3 ScoreTask 4 ScoreTask 5 ScoreOverall ScoreCRL2025.1091967289372TRL2025.1076968119257IQL2025.103182606336IVL2025.1029371603924BC2025.1010682117QRL2025.108223092018FBC2025.1078153129COE2025.1069131149TDP2025.1047120147