drawer-open on Meta-World v2 (test)
1,556Best Attack RewardPPO
Evaluation results
| Method | Links | |
|---|---|---|
| PPODefense mechanism=No defense2024.06 | 1,556 | |
| ATLA-PPODefense mechanism=AdvTraining2024.06 | 1,158 | |
| PA-ATLA-PPODefense mechanism=AdvTraining2024.06 | 954 | |
| RAD-PPODefense mechanism=Regret2024.06 | 736 | |
| WocaR-PPODefense mechanism=Partial smoothing2024.06 | 579 | |
| SA-PPODefense mechanism=Smoothing, Time-discounting=false2024.06 | 403 | |
| TDRT-PPODefense mechanism=Smoothing, Time-discounting=true2024.06 | 378 |