ResearchBenchmarksGenerative Recommendation on RecSim Medium Quality Mixed StrategyFollow0.374RewardDRPO0.215920.256960.2980.33904Feb 11, 2026Evaluation ResultsMethodMethodLinksRewardeCPMDRPO2026.020.3741.87DRPO-Exp2026.020.3721.86Adapt. BCmechanism=Adaptive fil...mechanism=Adaptive filtering2026.020.3671.84BPPO2026.020.3411.7BC2026.020.3381.69IQL2026.020.3371.68AWR2026.020.3331.66CRR2026.020.2221.11