Long-form Generation on DetailCaps 100 sampled instances
64.4ScoreInternVL3
Evaluation results
| Method | Links | |
|---|---|---|
| InternVL3Type=AR2026.05 | 64.4 | |
| LLaDA-V + OursType=Diff.2026.05 | 63.6 | |
| Qwen2.5Type=AR2026.05 | 63.3 | |
| LLaVA-OVType=AR2026.05 | 60.6 | |
| LLaVA-1.6Type=AR2026.05 | 60.6 | |
| LLaDA-VType=Diff.2026.05 | 59.8 | |
| LaViDa + OursType=Diff.2026.05 | 56.1 | |
| LaViDaType=Diff.2026.05 | 8.3 |