ResearchBenchmarksReinforcement Learning on AntMaze-total v2 (aggregate)Reinforcement Learning on AntMaze-total v2 (aggregate)Follow556.4Online Normalized Score (1M steps)DMG-22.256127.972278.2428.428Nov 12, 2024Evaluation resultsMethodMethodLinksOnline Normalized Score (1M steps)Offline Normalized ScoreDMGTraining Mode=offline-...Training Mode=offline-to-online fine-tuning2024.11556.4439.4IQLTraining Mode=offline-...Training Mode=offline-to-online fine-tuning2024.11490.4392.4TD3Training Mode=online t...Training Mode=online training from scratch2024.110—