ResearchBenchmarksReinforcement Learning on SAPIEN Humanoid + HopperReinforcement Learning on SAPIEN Humanoid + HopperFollow0.97IQMABD-NET0.41880.56190.7050.8481Mar 19, 2026Evaluation resultsMethodMethodLinksIQMMedian PerformanceMean PerformanceABD-NETTraining Algorithm=PPOTraining Algorithm=PPO2026.030.970.970.94GNNTraining Algorithm=PPOTraining Algorithm=PPO2026.030.730.760.64SWATTraining Algorithm=PPOTraining Algorithm=PPO2026.030.710.710.7BOTTraining Algorithm=PPOTraining Algorithm=PPO2026.030.660.670.59MLPTraining Algorithm=PPOTraining Algorithm=PPO2026.030.560.570.55RODRIGUESTraining Algorithm=PPOTraining Algorithm=PPO2026.030.440.450.41