Offline Reinforcement Learning on Walker2d v2 (offline)
0Observation ValueEDAC
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| EDACCorruption type=Adversarial, Corruption rate (c)=0.3, Corruption scale (epsilon)=1.02024.11 | 0 | 41.9 | 57.3 | 4.3 | |
| MSGCorruption type=Adversarial, Corruption rate (c)=0.3, Corruption scale (epsilon)=1.02024.11 | 2.9 | 5.4 | 9.6 | 0.1 | |
| UWMSGCorruption type=Adversarial, Corruption rate (c)=0.3, Corruption scale (epsilon)=1.02024.11 | 6.3 | 5.9 | 35.1 | 1.8 | |
| BCCorruption type=Adversarial, Corruption rate (c)=0.3, Corruption scale (epsilon)=1.02024.11 | 12.7 | 5.4 | 16 | 16 | |
| IQLCorruption type=Adversarial, Corruption rate (c)=0.3, Corruption scale (epsilon)=1.02024.11 | 37.7 | 27.5 | 73.5 | -0.1 | |
| CQLCorruption type=Adversarial, Corruption rate (c)=0.3, Corruption scale (epsilon)=1.02024.11 | 61.8 | 27 | 67 | 3.9 | |
| RIQLCorruption type=Adversarial, Corruption rate (c)=0.3, Corruption scale (epsilon)=1.02024.11 | 70 | 66.1 | 85 | 60.6 | |
| TRACERCorruption type=Adversarial, Corruption rate (c)=0.3, Corruption scale (epsilon)=1.02024.11 | 70 | 69.3 | 88.9 | 64 |