Goal-Conditioned Reinforcement Learning (Manipulation) on scene-play state-based v0 (test)
72Success RateDual Goal Representations
Evaluation Results
| Method | Links | |
|---|---|---|
| Dual Goal RepresentationsDownstream algorithm=GCIVL, Observation type=State-based2025.10 | 72 | |
| VIBDownstream algorithm=GCIVL, Observation type=State-based2025.10 | 58 | |
| OrigDownstream algorithm=GCIVL, Observation type=State-based2025.10 | 46 | |
| TRADownstream algorithm=GCIVL, Observation type=State-based2025.10 | 46 | |
| BYOL-γDownstream algorithm=GCIVL, Observation type=State-based2025.10 | 44 | |
| VIPDownstream algorithm=GCIVL, Observation type=State-based2025.10 | 23 |