ResearchBenchmarksReinforcement Learning on AntMaze medium-diverse v2Reinforcement Learning on AntMaze medium-diverse v2Follow96.2Online Score (1M steps)DMG-3.84822.12648.174.074Nov 12, 2024Evaluation resultsMethodMethodLinksOnline Score (1M steps)Offline ScoreDMGTraining Mode=offline-...Training Mode=offline-to-online fine-tuning2024.1196.277.2IQLTraining Mode=offline-...Training Mode=offline-to-online fine-tuning2024.1190.272.8TD3Training Mode=online t...Training Mode=online training from scratch2024.110—