Goal-conditioned Reinforcement Learning (G1_speed) on walker2d medium v2
2.904Goal-conditioned ReturnPoE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PoE2026.04 | 2.904 | 15.234 | |
| KL-Reg2026.04 | 2.9 | 14.795 | |
| Prior Only2026.04 | 2.887 | 87.928 | |
| Additive2026.04 | 2.826 | 21.892 | |
| Frozen2026.04 | 2.562 | 0 |