Offline Goal-conditioned Reinforcement Learning on D4RL AntMaze umaze v2
87.5Normalized ScoreIQL*
Evaluation Results
| Method | Links | |
|---|---|---|
| IQL*TD Learning=true2022.06 | 87.5 | |
| Contrastive RL + BCTD Learning=false, Number of networks=22022.06 | 81.9 | |
| Contrastive RL + BCTD Learning=false, Number of networks=52022.06 | 79.8 | |
| TD3+BC*TD Learning=true2022.06 | 78.6 | |
| DTTD Learning=false2022.06 | 65.6 | |
| GCBCTD Learning=false2022.06 | 65.4 | |
| BCTD Learning=false2022.06 | 54.6 |