Reinforcement Learning on OGBench Offline
94AL ScoreReBRAC
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ReBRACPolicy Extraction Mode=GAUSSIAN2026.06 | 94 | 57 | 69 | 17 | 65 | 79 | 0 | 9 | 1 | 9 | 40 | |
| QAM-EPolicy Extraction Mode=FINE-TUNING2026.06 | 83 | 1 | 59 | 2 | 97 | 100 | 39 | 65 | 5 | 6 | 46 | |
| QPILOTS-UPolicy Extraction Mode=OURS2026.06 | 82 | 1 | 63 | 8 | 89 | 100 | 16 | 75 | 7 | 3 | 44 | |
| QPILOTS-MPolicy Extraction Mode=OURS2026.06 | 82 | 10 | 71 | 12 | 95 | 100 | 17 | 75 | 7 | 10 | 48 | |
| QAMPolicy Extraction Mode=FINE-TUNING2026.06 | 81 | 18 | 67 | 11 | 97 | 99 | 0 | 64 | 3 | 3 | 44 | |
| FQLPolicy Extraction Mode=FINE-TUNING2026.06 | 76 | 0 | 68 | 9 | 79 | 70 | 5 | 46 | 3 | 2 | 36 | |
| CGQLPolicy Extraction Mode=STEERING2026.06 | 76 | 0 | 60 | 5 | 38 | 48 | 24 | 38 | 8 | 0 | 30 | |
| CGQL-MPolicy Extraction Mode=STEERING2026.06 | 71 | 4 | 42 | 6 | 74 | 100 | 0 | 41 | 8 | 1 | 35 | |
| CGQL-LPolicy Extraction Mode=STEERING2026.06 | 65 | 3 | 62 | 6 | 88 | 90 | 0 | 45 | 8 | 0 | 37 | |
| DSRLPolicy Extraction Mode=STEERING2026.06 | 61 | 3 | 53 | 3 | 99 | 87 | 0 | 74 | 1 | 2 | 38 | |
| FEditPolicy Extraction Mode=STEERING2026.06 | 58 | 2 | 22 | 3 | 62 | 99 | 34 | 40 | 2 | 5 | 33 | |
| IFQLPolicy Extraction Mode=STEERING2026.06 | 36 | 1 | 86 | 24 | 84 | 100 | 0 | 11 | 0 | 2 | 34 | |
| FBRACPolicy Extraction Mode=BACKPROP2026.06 | 2 | 0 | 39 | 0 | 50 | 0 | 15 | 0 | 0 | 0 | 11 |