ResearchBenchmarksConstrained Offline Reinforcement Learning on DSRL CarGoal1Follow0.79Normalized ReturnCPQ0.2180.36650.5150.6635Dec 9, 2025Evaluation ResultsMethodMethodLinksNormalized ReturnCostCPQ2025.120.791.42CDT2025.120.661.21MPDiffusernum_samples=16num_samples=162025.120.630.92BCQ-Lag2025.120.470.78BC-All2025.120.390.33COptiDICE2025.120.350.54BC-Safe2025.120.240.28