Locomotion Control on D4RL Walker2d medium-expert
111.2Normalized ReturnRepb-SDE
Evaluation Results
| Method | Links | |
|---|---|---|
| Repb-SDEalgorithm class=model-based2021.10 | 111.2 | |
| ROMI-CQLbase algorithm=CQL2021.10 | 109.7 | |
| CQLalgorithm class=model-free2021.10 | 109.5 | |
| Fisher-BRCimplementation=author-provided (re-run)2021.06 | 103.6 | |
| COMBO2021.02 | 103.3 | |
| TD3+BCalpha=2.52021.06 | 101.1 | |
| KIOAmount of data used=100k, Evaluation=Average score over 100 episodes2024.10 | 100.1 | |
| CQL2021.02 | 98.7 | |
| MOREL2021.02 | 95.6 | |
| TeacherEvaluation=Average score over 100 episodes2024.10 | 82.7 | |
| BRAC-v2021.02 | 81.6 | |
| BRAC-p2021.02 | 76.9 | |
| BC2021.10 | 73.4 | |
| CQLimplementation=author-provided (re-run)2021.06 | 68.1 | |
| MOPO2021.02 | 44.6 | |
| AWACimplementation=reported2021.06 | 42.7 | |
| BEAR2021.02 | 40.1 | |
| BC(TD3+BC)Evaluation=Average score over 100 episodes2024.10 | 12 | |
| BCimplementation=own2021.06 | 12 | |
| BC(CQL)Evaluation=Average score over 100 episodes2024.10 | 11.3 | |
| BC2021.02 | 6.4 | |
| SAC-off2021.02 | -0.1 | |
| BRAC-pimplementation=reported2021.06 | -0.3 |