ResearchBenchmarksMulti-task Reinforcement Learning on Meta-World MT50-rand V2 (Sub-optimal)Follow48.94Average Success RateMTDIFF-P-ONEHOT19.497627.141334.78542.4287May 29, 2023Evaluation ResultsMethodMethodLinksAverage Success RateMTDIFF-P-ONEHOTTraining Protocol=OfflineTraining Protocol=Offline2023.0548.94MTDIFF-PTraining Protocol=OfflineTraining Protocol=Offline2023.0548.67MTIQLTraining Protocol=OfflineTraining Protocol=Offline2023.0543.28PromptDTTraining Protocol=OfflineTraining Protocol=Offline2023.0539.76MTBCTraining Protocol=OfflineTraining Protocol=Offline2023.0534.53MTDTTraining Protocol=OfflineTraining Protocol=Offline2023.0520.63