Offline Reinforcement Learning on ExORL
104.7Cheetah Run ScoreReBRAC
Evaluation results
| Method | Links | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ReBRACAugmentation=MBPO2025.06 | 104.7 | 87 | 447.4 | 398.6 | 327.6 | 107.6 | 458.1 | 341.8 | 395.1 | 666.9 | 703.7 | 390 | 60.1 | 52.5 | 43.4 | 34 | |
| InFOM2025.06 | 97.6 | 104.7 | 254.8 | 251.8 | 467.3 | 116.3 | 581.2 | 358.8 | 341.8 | 626 | 718.3 | 360.7 | 742.5 | 687.5 | 746.7 | 733 | |
| ReBRAC2025.06 | 97.2 | 84.9 | 443.4 | 403 | 208.1 | 97.8 | 460.6 | 344.6 | 343 | 605.2 | 688.6 | 343.5 | 59 | 38 | 44.5 | 41.4 | |
| IQLModel=HILP2025.06 | 90.3 | 64.4 | 366.8 | 318.1 | 225.4 | 97.4 | 443.3 | 280.7 | 371.1 | 626.5 | 715.6 | 393.4 | 40.4 | 25.1 | 27.8 | 24.7 | |
| ReBRACEncoder=DINO2025.06 | 87.2 | 67.1 | 383.5 | 318.4 | 228 | 98.5 | 453 | 320.3 | 344.7 | 573 | 663.2 | 391.4 | 17.5 | 11 | 13.7 | 8.3 | |
| IQLModel=FB2025.06 | 83.3 | 67.3 | 346.5 | 298 | 225.3 | 97.4 | 446.8 | 287 | 367 | 639.4 | 728.9 | 388.9 | 25 | 16.2 | 19.8 | 12.4 | |
| IQL2025.06 | 80 | 77 | 357.9 | 303.7 | 208.6 | 92.4 | 409.1 | 260.3 | 358 | 628.5 | 714.2 | 375.1 | 74.7 | 40.6 | 77.1 | 78.7 | |
| CRLConditioning=IS2025.06 | 73.3 | 74.7 | 327.4 | 330.2 | 213.3 | 91.5 | 409 | 255 | 323.4 | 576.7 | 653.1 | 309.6 | 29.1 | 21.4 | 19.8 | 19.6 | |
| TD InfoNCEConditioning=IS2025.06 | 68.2 | 74.3 | 336.7 | 326.3 | 212.2 | 91 | 397.2 | 231.6 | 222.1 | 421.4 | 457.1 | 243.1 | 31.5 | 25.5 | 26.6 | 25.4 |