ResearchBenchmarksConstrained Offline Reinforcement Learning on DSRL Walker2dVelocityFollow0.8Normalized ReturnMPDiffuser0.00960.21480.420.6252Dec 9, 2025Evaluation ResultsMethodMethodLinksNormalized ReturnCostMPDiffusernum_samples=16num_samples=162025.120.80.27BC-All2025.120.793.88BC-Safe2025.120.790.04BCQ-Lag2025.120.790.17CDT2025.120.780.06COptiDICE2025.120.120.74CPQ2025.120.040.21