ResearchBenchmarksConstrained Offline Reinforcement Learning on DSRL HopperVelocityFollow0.81Normalized ReturnMPDiffuser0.10280.28640.470.6536Dec 9, 2025Evaluation ResultsMethodMethodLinksNormalized ReturnCostMPDiffusernum_samples=16num_samples=162025.120.810.37BCQ-Lag2025.120.785.02BC-All2025.120.656.39CDT2025.120.630.61BC-Safe2025.120.360.67CPQ2025.120.142.11COptiDICE2025.120.131.51