Continuous Control on walker2d
5,111Avg RewardTOP-TD3
Evaluation results
| Method | Links | ||
|---|---|---|---|
| TOP-TD3Training steps=1M, Number of trials=102021.02 | 5,111 | — | |
| SACTraining steps=1M, Number of trials=102021.02 | 4,984 | — | |
| ND TOP-TD3Training steps=1M, Number of trials=102021.02 | 4,832 | — | |
| OACTraining steps=1M, Number of trials=102021.02 | 4,786 | — | |
| QR-TD3Training steps=1M, Number of trials=102021.02 | 4,560 | — | |
| TD3Training steps=1M, Number of trials=102021.02 | 4,412 | — | |
| DDPGObservation Space=pixels2015.09 | 0.944 | 1.476 | |
| DDPGObservation Space=low-dimensional2015.09 | 0.705 | 1.573 | |
| DPGSetup=original with replay buffer and batch normalization2015.09 | 0.393 | 1.397 |