ResearchBenchmarksOffline Goal-conditioned Reinforcement Learning on AntMaze Large Navigate (oraclerep v0)Follow87Task 1 ScoreCRL-0.3622.324567.68Oct 26, 2025Evaluation ResultsMethodMethodLinksTask 1 ScoreTask 2 ScoreTask 3 ScoreTask 4 ScoreTask 5 ScoreOverall ScoreCRL2025.10876491929085TRL2025.1057668081846QRL2025.10516996576167IQL2025.10234482152037FBC2025.1013244871622IVL2025.1011195541821TDP2025.1062369171727COE2025.1062667141826BC2025.1032253171722