ResearchBenchmarksReinforcement Learning on MuJoCo Ant (Robustness Metrics)Follow4,720,255Clean RewardPPO4,250,265.484,372,281.994,494,298.54,616,315.01Jun 6, 2024Evaluation ResultsMethodMethodLinksClean RewardBest Attack RewardPPOSmoothing=false, Time-...Smoothing=false, Time-discounting=false2024.064,720,255850,612TDRT-PPOSmoothing=true, Time-d...Smoothing=true, Time-discounting=true2024.064,612,2193,206,781SA-PPOSmoothing=true, Time-d...Smoothing=true, Time-discounting=false2024.064,268,3423,318,391