Visual Grounding on GQA Lite
75.8AccuracyLLaVA-1.6
Evaluation Results
| Method | Links | |
|---|---|---|
| LLaVA-1.6Type=AR2026.05 | 75.8 | |
| LLaVA-OVType=AR2026.05 | 72.2 | |
| InternVL3Type=AR2026.05 | 71.2 | |
| Qwen2.5Type=AR2026.05 | 70.6 | |
| LLaDA-VType=Diff.2026.05 | 61.6 | |
| LLaDA-V + OursType=Diff.2026.05 | 61.6 | |
| LaViDa + OursType=Diff.2026.05 | 60.2 | |
| LaViDaType=Diff.2026.05 | 59.4 |