ResearchBenchmarksOffline Reinforcement Learning on Maze2D Single TaskFollow124.8U-Maze ScoreIMLE0.93633.09365.2597.407Mar 14, 2026Evaluation ResultsMethodMethodLinksU-Maze ScoreMedium Maze ScoreLarge Maze ScoreSingle-task Average ScoreIMLEBatch Size=1Batch Size=12026.03124.8117.3129.2123.7DiffuserBatch Size=1Batch Size=12026.03113.9121.5123119.5IQLBatch Size=1Batch Size=12026.0347.434.958.647MPPIBatch Size=1Batch Size=12026.0333.210.25.116.2CQLBatch Size=1Batch Size=12026.035.7512.57.7