Visual Question Answering on VQA-AD
50AccuracyOffline ICD
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Offline ICDBackbone=LLaVA-OneVision2025.10 | 50 | 100 | |
| ICDBackbone=LLaVA-OneVision2025.10 | 47 | 18.2 | |
| BestofNBackbone=LLaVA-OneVision2025.10 | 42 | — | |
| Self-labelingBackbone=LLaVA-OneVision2025.10 | 41 | — | |
| GPT-4o2025.10 | 41 | — | |
| 0-shotBackbone=LLaVA-OneVision2025.10 | 37 | — | |
| Offline ICDBackbone=InternVL2.52025.10 | 37 | 100 | |
| ICDBackbone=InternVL2.52025.10 | 33 | 14.5 | |
| CoTBackbone=LLaVA-OneVision2025.10 | 28 | — | |
| CoTBackbone=InternVL2.52025.10 | 24 | — | |
| BestofNBackbone=InternVL2.52025.10 | 23 | — | |
| Self-labelingBackbone=InternVL2.52025.10 | 21 | — | |
| 0-shotBackbone=InternVL2.52025.10 | 19 | — |