faucet-close on Meta-World v2 (test)
4,108Best Attack RewardATLA-PPO
Evaluation Results
| Method | Links | |
|---|---|---|
| ATLA-PPODefense mechanism=AdvTraining2024.06 | 4,108 | |
| PA-ATLA-PPODefense mechanism=AdvTraining2024.06 | 4,012 | |
| PPODefense mechanism=No defense2024.06 | 3,409 | |
| WocaR-PPODefense mechanism=Partial smoothing2024.06 | 2,829 | |
| RAD-PPODefense mechanism=Regret2024.06 | 2,235 | |
| TDRT-PPODefense mechanism=Smoothing, Time-discounting=true2024.06 | 1,789 | |
| SA-PPODefense mechanism=Smoothing, Time-discounting=false2024.06 | 1,559 |