Image Quality Evaluation on Regular Dataset Complex Vision-4B
85.8Overall ScoreDPO
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| DPOModel Scale=4B, Visual Grounding=True, Short Name=D2026.03 | 85.8 | 88.31 | 83 | 85.62 | 83.69 | 87.62 | 86.62 | |
| Reward-WeightedModel Scale=4B, Visual Grounding=True, Short Name=RW2026.03 | 84.98 | 87.85 | 82.46 | 84.92 | 82.54 | 86.92 | 85.15 | |
| BModel Scale=4B, Visual Grounding=True2026.03 | 84.84 | 86.69 | 81.92 | 84.69 | 82.69 | 87.08 | 86 | |
| Standardized Reward-WeightedModel Scale=4B, Visual Grounding=True, Short Name=SW2026.03 | 84.8 | 87.77 | 82.62 | 85.62 | 82.46 | 86.85 | 83.54 | |
| RModel Scale=4B, Visual Grounding=True2026.03 | 84.78 | 87.08 | 82.77 | 85.62 | 82.54 | 86.62 | 84.08 | |
| EModel Scale=4B, Visual Grounding=True2026.03 | 83.94 | — | 85.46 | — | — | 84.38 | — | |
| SModel Scale=4B, Visual Grounding=True2026.03 | 82.8 | 84.92 | 79.85 | 82.69 | 82.54 | 85.31 | 84 | |
| GPT-4o PlannerModel Scale=4B, Visual Grounding=True, Short Name=G4o2026.03 | 77.15 | 78.38 | 76.08 | 77.85 | 74.85 | 76.77 | 79.15 |