Visual Grounding on RefCOCO
66.9ASR@0.5IAG
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| IAGModel=InternVL-2.5-8B2025.08 | 66.9 | — | 89.5 | — | |
| ImperioModel=InternVL-2.5-8B2025.08 | 65.5 | — | 89.2 | — | |
| IAGModel=Llava-v1.5-7B2025.08 | 58.9 | — | 80.7 | — | |
| ImperioModel=Llava-v1.5-7B2025.08 | 55.2 | — | 80.5 | — | |
| IAGModel=Ferret-7B2025.08 | 48.9 | — | 85.3 | — | |
| ImperioModel=Ferret-7B2025.08 | 35.6 | — | 80.9 | — | |
| MarksmanModel=Ferret-7B2025.08 | 33.4 | — | 85.9 | — | |
| MarksmanModel=InternVL-2.5-8B2025.08 | 16.4 | — | 89.7 | — | |
| MarksmanModel=Llava-v1.5-7B2025.08 | 8.5 | — | 84.8 | — | |
| RandomModel=InternVL-2.5-8B2025.08 | 8.1 | — | — | 90.3 | |
| One-to-NModel=Ferret-7B2025.08 | 4.9 | — | 85.8 | — | |
| RandomModel=Ferret-7B2025.08 | 4.7 | — | — | 87.5 | |
| One-to-NModel=InternVL-2.5-8B2025.08 | 3.8 | — | 89.5 | — | |
| One-to-NModel=Llava-v1.5-7B2025.08 | 3.2 | — | 76.5 | — | |
| RandomModel=Llava-v1.5-7B2025.08 | 2 | — | — | 82.1 | |
| LG-VQCodebook=MS-COCO2024.05 | — | 9.62 | — | — | |
| VQ-GANCodebook=MS-COCO2024.05 | — | 9.14 | — | — | |
| VQCTCodebook=MS-COCO2024.05 | — | 9.46 | — | — |