ResearchBenchmarksManipulation on D4RL Adroit pen-humanFollow106.2Normalized ScoreCSVE0.43227.89155.3582.809May 14, 2026May 16, 2026May 19, 2026May 22, 2026May 25, 2026May 28, 2026May 31, 2026Evaluation ResultsMethodMethodLinksNormalized ScoreCSVE2026.05106.2MISA2026.0588.1DPPO2026.0576.3CPQL2026.0572.1IQL2026.0571.5FB-PbRLfine-tuning=truefine-tuning=true2026.0570.2MCQ2026.0568.5EPQreproduced=truereproduced=true2026.0565.7TD3+BC2026.0564.8CQL2026.0537.5BCreproduced=truereproduced=true2026.0534.4LiRE2026.054.5