Continuous Control on DMC-GB video hard
54,443Cartpole Swingup ScoreSGQN
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| SGQNzero-shot=true, training steps=500,0002022.09 | 54,443 | 85,124 | 73,921 | 78,257 | — | 82,224 | 748 | 28,060 | |
| SODAzero-shot=true, training steps=500,0002022.09 | 41,317 | 736,132 | 31,232 | 381,163 | — | 30,949 | 430 | — | |
| SVEAzero-shot=true, training steps=500,0002022.09 | 40,138 | 74,743 | 38,563 | 498,174 | — | 30,724 | 468 | — | |
| SACzero-shot=true, training steps=500,0002022.09 | 15,817 | 23,157 | 12,247 | 10,137 | — | 1,310 | 125 | — | |
| RADzero-shot=true, training steps=500,0002022.09 | 15,229 | 22,945 | 8,010 | 9,840 | — | 3,920 | 119 | — | |
| DrQzero-shot=true, training steps=500,0002022.09 | 13,629 | 25,257 | 12,152 | 10,040 | — | 3,813 | 129 | — | |
| SMG2024.09 | 764 | 955 | 814 | — | 303 | 910 | — | — | |
| SGQN2024.09 | 599 | 870 | 634 | — | 135 | 710 | — | — | |
| SVEABackground Mode=overlay2024.09 | 510 | 814 | 348 | — | 105 | 353 | — | — | |
| PIE-GGeneralization=Zero-shot2022.12 | 401 | 852 | 600 | 786 | 154 | 762 | — | — | |
| SVEAGeneralization=Zero-shot2022.12 | 393 | 834 | 377 | 403 | 105 | 335 | — | — | |
| SODA2024.09 | 346 | 406 | 175 | — | 118 | 310 | — | — | |
| TLDAGeneralization=Zero-shot2022.12 | 286 | 602 | 271 | 257 | 90 | 241 | — | — | |
| SRM2024.09 | 254 | 558 | 165 | — | 87 | 131 | — | — | |
| DrQ2024.09 | 168 | 278 | 110 | — | 38 | 54 | — | — | |
| SAC2024.09 | 156 | 212 | 132 | — | 56 | 22 | — | — | |
| DrQGeneralization=Zero-shot2022.12 | 138 | 289 | 104 | 92 | 32 | 71 | — | — | |
| DrQ-v2Generalization=Zero-shot2022.12 | 130 | 151 | 34 | 97 | 23 | 21 | — | — | |
| ACO-MoE2026.04 | — | — | — | — | — | — | 809 | — | |
| FTR†2026.04 | — | — | — | — | — | — | 762 | — | |
| SGQN2026.04 | — | — | — | — | — | — | 590 | — | |
| SMG2026.04 | — | — | — | — | — | — | 749 | — |