Referring Expression Grounding on RefCOCO
92.64AccuracySingle Task SoTA
Evaluation results
| Method | Links | |
|---|---|---|
| Single Task SoTA2023.12 | 92.64 | |
| UIO-2XXLModel Size=6.8B2023.12 | 90.7 | |
| Qwen-VLModel Size=9.6B2023.12 | 89.4 | |
| UIO-2XLModel Size=3.2B2023.12 | 88.2 | |
| FerretModel Size=7.2B2023.12 | 87.5 | |
| ShikraModel Size=7.2B2023.12 | 87 | |
| UIO-2LModel Size=1.1B2023.12 | 84.1 | |
| SWUDImode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 76.24 | |
| SWUDI-Amode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 76.2 | |
| OptMergemode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 76.09 | |
| WUDI Mergingmode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 76.06 | |
| TIES Mergingmode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 76.01 | |
| TIES w/ DAREmode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 75.1 | |
| TA w/ DAREmode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 74.94 | |
| Task Arithmeticmode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 74.9 | |
| Iso-Cmode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 72.72 | |
| Mixture Trainingmode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 72.06 | |
| InternVL2.5-Instructmode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 71.69 | |
| TSV Mergingmode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 65.19 | |
| Weight Averagemode=Full fine-tuning, backbone=InternVL2.5-1B2026.06 | 52.62 |