ResearchBenchmarksOffline Reinforcement Learning on Maze2D Multi TaskOffline Reinforcement Learning on Maze2D Multi TaskFollow132.3U-Maze ScoreIMLE20.549.52578.55107.575Mar 14, 2026Evaluation resultsMethodMethodLinksU-Maze ScoreMedium Maze ScoreLarge Maze ScoreAverage Score (Multi-task)IMLEBatch Size=1Batch Size=12026.03132.3127.8137.1132.4DiffuserBatch Size=1Batch Size=12026.03128.9127.2132.1129.4MPPIBatch Size=1Batch Size=12026.0341.215.4821.5IQLBatch Size=1Batch Size=12026.0324.812.113.916.9