Goal-Conditioned Reinforcement Learning on manipulation-cube-single-play (test)
0.9Success RateVIB
Evaluation Results
| Method | Links | |
|---|---|---|
| VIBDownstream algorithm=GCIVL, Observation type=State-based2025.10 | 0.9 | |
| Dual Goal RepresentationsDownstream algorithm=GCIVL, Observation type=State-based2025.10 | 0.89 | |
| OrigDownstream algorithm=GCIVL, Observation type=State-based2025.10 | 0.52 | |
| BYOL-γDownstream algorithm=GCIVL, Observation type=State-based2025.10 | 0.51 | |
| VIPDownstream algorithm=GCIVL, Observation type=State-based2025.10 | 0.4 | |
| TRADownstream algorithm=GCIVL, Observation type=State-based2025.10 | 0.4 | |
| CRL2025.12 | 0.32 | |
| HIQL2025.12 | 0.31 | |
| Eik-HIQL2025.12 | 0.25 | |
| Eik-HiQRL2025.12 | 0.12 | |
| QRL2025.12 | 0.11 |