Offline Goal-conditioned Reinforcement Learning on D4RL AntMaze large-diverse v2
54.1Normalized ScoreContrastive RL + BC
Evaluation Results
| Method | Links | |
|---|---|---|
| Contrastive RL + BCTD Learning=false, Number of networks=52022.06 | 54.1 | |
| Contrastive RL + BCTD Learning=false, Number of networks=22022.06 | 49.3 | |
| IQL*TD Learning=true2022.06 | 47.5 | |
| GCBCTD Learning=false2022.06 | 36.9 | |
| DTTD Learning=false2022.06 | 0.2 | |
| BCTD Learning=false2022.06 | 0 | |
| TD3+BC*TD Learning=true2022.06 | 0 |