Offline Reinforcement Learning on DeepMind Control Suite hopper-stand medium
65Mean Normalized ReturnLAWM
Evaluation Results
| Method | Links | |
|---|---|---|
| LAWMaction-conditioned data percentage=5%2025.12 | 65 | |
| C-LAPaction-conditioned data percentage=5%2025.12 | 60 | |
| C-LAPaction-conditioned data percentage=100%, variant=Oracle2025.12 | 59.5 | |
| TD3+BCaction-conditioned data percentage=100%, variant=Oracle2025.12 | 45.8 | |
| IDM-TD3+BCaction-conditioned data percentage=5%2025.12 | 41.4 | |
| TD3+BCaction-conditioned data percentage=5%2025.12 | 39.6 |