Visual Grounding on RefCOCO Series
77.73RefCOCO Test ScoreVanilla
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| VanillaBase Model=LLaVA-Next 7B2026.02 | 77.73 | 76.34 | 57.25 | 71.05 | 100 | — | |
| NüwaBase Model=LLaVA-Next 7B, Average Tokens=6402026.02 | 68.17 | 66.91 | 49.54 | 59.72 | 86.48 | — | |
| NüwaBase Model=LLaVA-Next 7B, Average Tokens=3202026.02 | 38.67 | 33.86 | 25.36 | 31.49 | 45.68 | — | |
| NüwaBase Model=LLaVA-Next 7B, Average Tokens=1602026.02 | 18.67 | 15.44 | 10.45 | 17.04 | 21.62 | — | |
| NuwaBackbone=QWEN-2.5-VL 7B, Average Tokens=75%2026.02 | — | — | — | — | — | 98.8 | |
| NuwaBackbone=QWEN-2.5-VL 7B, Average Tokens=50%2026.02 | — | — | — | — | — | 96.4 | |
| NuwaBackbone=QWEN-2.5-VL 7B, Average Tokens=25%2026.02 | — | — | — | — | — | 83.8 | |
| VanillaBackbone=QWEN-2.5-VL 7B, Average Tokens=100%2026.02 | — | — | — | — | — | 100 |