Offline Reinforcement Learning on Halfcheetah
7,357.5Average ReturnSAC-N
Evaluation results
| Method | Links | |
|---|---|---|
| SAC-NAttack Type=Random Reward2023.10 | 7,357.5 | |
| UWMSGAttack Type=Random Reward2023.10 | 7,299.9 | |
| EDACAttack Type=Random Reward2023.10 | 7,128.2 | |
| MSGAttack Type=Random Reward2023.10 | 4,339.5 | |
| UWMSGAttack Type=Adversarial Dynamics2023.10 | 4,144.3 | |
| UWMSGAttack Type=Random Dynamics2023.10 | 1,425 | |
| UWMSGAttack Type=Adversarial Reward2023.10 | 1,016.7 | |
| EDACAttack Type=Adversarial Dynamics2023.10 | 374 | |
| MSGAttack Type=Adversarial Reward2023.10 | 243.1 | |
| MSGAttack Type=Random Dynamics2023.10 | 212.9 | |
| EDACAttack Type=Random Dynamics2023.10 | -12.3 | |
| SAC-NAttack Type=Adversarial Reward2023.10 | -55.7 | |
| SAC-NAttack Type=Random Dynamics2023.10 | -66.2 | |
| MSGAttack Type=Adversarial Dynamics2023.10 | -87.3 | |
| EDACAttack Type=Adversarial Reward2023.10 | -127.2 | |
| SAC-NAttack Type=Adversarial Dynamics2023.10 | -246.8 |