Image Styling on Complex
85.41Overall ScoreDPO (D)
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| DPO (D)Model Size=8B, Modality=Vision2026.03 | 85.41 | 87.12 | 82.97 | 85.51 | 83.98 | 87.03 | 85.68 | |
| RModel Scale=8B2026.03 | 84.46 | 87.39 | 81.67 | 84.71 | 82.39 | 86.3 | 84.28 | |
| RWModel Scale=8B, Training strategy=Reward-Weighted2026.03 | 84.44 | 86.59 | 83.04 | 85.29 | 82.9 | 85.65 | 83.19 | |
| EModel Scale=8B2026.03 | 84.08 | — | 85.65 | — | — | 84.71 | — | |
| BModel Scale=8B2026.03 | 84.06 | 86.67 | 82.25 | 83.84 | 82.32 | 85.72 | 83.62 | |
| DModel Scale=8B, Method Identity=DPO2026.03 | 83.94 | 87.1 | 81.59 | 85.94 | 81.88 | 85.36 | 81.81 | |
| SWModel Scale=8B, Training strategy=Standardized Reward-Weighted2026.03 | 83.83 | 85.72 | 81.38 | 85.22 | 82.03 | 84.57 | 83.91 | |
| Standardized Reward-Weighted (SW)Model Size=8B, Modality=Vision2026.03 | 83.6 | 86.69 | 81.78 | 83.98 | 81.61 | 84.92 | 82.54 | |
| Reward-Weighted (RW)Model Size=8B, Modality=Vision2026.03 | 83.55 | 86.69 | 81.27 | 84.07 | 80.85 | 85.93 | 82.63 | |
| Edit-Only (E)Model Size=8B, Modality=Vision2026.03 | 83.38 | — | 84.07 | — | — | 83.81 | — | |
| Baseline (B)Model Size=8B, Modality=Vision2026.03 | 82.96 | 85.08 | 80.85 | 83.22 | 80.51 | 84.92 | 83.31 | |
| SModel Scale=8B2026.03 | 82.93 | 85.94 | 81.01 | 83.55 | 81.01 | 84.28 | 81.74 | |
| Method RModel Size=8B, Modality=Vision2026.03 | 82.9 | 85.59 | 80.93 | 83.9 | 80.51 | 84.24 | 82.2 | |
| Method SModel Size=8B, Modality=Vision2026.03 | 82.61 | 85.17 | 80.42 | 82.71 | 80.17 | 84.32 | 82.71 | |
| G4oMethod Identity=GPT-4o Planner, Config=zero-shot2026.03 | 78.01 | 79.64 | 77.97 | 79.78 | 76.3 | 76.59 | 77.83 | |
| GPT-4o (G4o)Model Size=Closed-source, Zero-shot=true2026.03 | 77.66 | 78.39 | 77.12 | 79.15 | 75.34 | 77.37 | 78.73 |