Edit Inspectors Question Answering on MagicBrush (test)
73.7AccuracyGPT-4
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| GPT-42025.06 | 73.7 | 48.2 | 49.2 | 51.6 | 63.2 | |
| GPT-4o2025.06 | 71 | 56.7 | 47.6 | 60.6 | 60.7 | |
| Qwen2.5 VL2025.06 | 64 | 38.4 | 47.5 | 50 | 59.8 | |
| InternVL32025.06 | 63.6 | 45.6 | 42.6 | 52.5 | 58.4 | |
| LLaVAtraining=Supervised2025.06 | 62.3 | — | — | 56.2 | 46.5 | |
| GPT-4 Turbo2025.06 | 60 | 59.7 | 49.8 | 54.1 | 61.1 | |
| LLaVA2025.06 | 58.5 | — | — | 44.8 | 50 |