ResearchBenchmarksOffline Reinforcement Learning on Hopper Medium-Replay Noise 2.5Follow81.33Normalized ReturnI-TAP20.479636.277352.07567.8727Feb 21, 2026Evaluation ResultsMethodMethodLinksNormalized ReturnI-TAPModel Category=Model-B...Model Category=Model-Based2026.0281.33L-MAPModel Category=Model-B...Model Category=Model-Based2026.0273.81IQLModel Category=Model-FreeModel Category=Model-Free2026.0249.12CQLModel Category=Model-FreeModel Category=Model-Free2026.0240.53TAPModel Category=Model-B...Model Category=Model-Based2026.0238.1DTModel Category=Model-FreeModel Category=Model-Free2026.0235.761R2RModel Category=Model-B...Model Category=Model-Based2026.0222.82