Offline Reinforcement Learning on DeepMind Control Suite hopper-stand expert
72.1Mean Normalized ReturnTD3+BC
Evaluation Results
| Method | Links | |
|---|---|---|
| TD3+BCaction-conditioned data percentage=100%, variant=Oracle2025.12 | 72.1 | |
| IDM-TD3+BCaction-conditioned data percentage=5%2025.12 | 56.5 | |
| TD3+BCaction-conditioned data percentage=5%2025.12 | 46.5 | |
| C-LAPaction-conditioned data percentage=100%, variant=Oracle2025.12 | 40 | |
| LAWMaction-conditioned data percentage=5%2025.12 | 37 | |
| C-LAPaction-conditioned data percentage=5%2025.12 | 13.6 |