Goal-conditioned Reinforcement Learning on OGBench visual cube single play (5 tasks) zero-shot
14Average ReturnOne-Step FB
Evaluation Results
| Method | Links | |
|---|---|---|
| One-Step FBevaluation_mode=Zero-shot evaluation2026.02 | 14 | |
| FBevaluation_mode=Zero-shot evaluation2026.02 | 12 | |
| BYOL-gammaevaluation_mode=Zero-shot evaluation2026.02 | 11 | |
| HILPevaluation_mode=Zero-shot evaluation2026.02 | 8 |