Reinforcement Learning on DMControl Finger Spin (OOD)
242.8Score (Friction 0.5)HaM-World
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| HaM-Worldprotocol=zero-shot2026.05 | 242.8 | 253.1 | 232.6 | 211.2 | 74.4 | 91.5 | 184.3 | |
| TD-MPC2protocol=zero-shot2026.05 | 227.9 | 227.7 | 220.2 | 196.3 | 47.9 | 84 | 167.3 | |
| SACprotocol=zero-shot2026.05 | 152.6 | 175.8 | 141.1 | 129.5 | 58 | 53.2 | 118.4 | |
| PPOprotocol=zero-shot2026.05 | 67.4 | 75.3 | 63.1 | 58.3 | 6.3 | 13.5 | 47.3 | |
| DreamerV3protocol=zero-shot2026.05 | 65 | 75.4 | 59.7 | 58.7 | 26.2 | 45.9 | 55.2 |