ResearchBenchmarksConstrained Offline Reinforcement Learning on DSRL HalfCheetahVelocityFollow105Normalized ReturnBCQ-Lag25.9646.486787.52Dec 9, 2025Evaluation ResultsMethodMethodLinksNormalized ReturnCostBCQ-Lag2025.1210518.21CDT2025.121000.01MPDiffusernum_samples=16num_samples=162025.12980.77BC-All2025.129713.1BC-Safe2025.12880.54COptiDICE2025.12650CPQ2025.12290.74