Goal-conditioned Reinforcement Learning on OGBench cube single play (5 tasks) zero-shot
30Average ReturnHILP
Evaluation Results
| Method | Links | |
|---|---|---|
| HILPevaluation_mode=Zero-shot evaluation2026.02 | 30 | |
| BYOL-gammaevaluation_mode=Zero-shot evaluation2026.02 | 13 | |
| ICVFevaluation_mode=Zero-shot evaluation2026.02 | 13 | |
| Laplacianevaluation_mode=Zero-shot evaluation2026.02 | 6 | |
| One-Step FBevaluation_mode=Zero-shot evaluation2026.02 | 3 | |
| FBevaluation_mode=Zero-shot evaluation2026.02 | 2 |