Visual Question Answering on UDK-VQA 1.0 (test)
90.6Score (Political)SearchLVLMs
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| SearchLVLMsbase_model=InternVL-1.52024.05 | 90.6 | 95.4 | 98 | 88.9 | 100 | 96 | 98.5 | 92.9 | |
| SearchLVLMsbase_model=GPT-4o2024.05 | 86.7 | 89.6 | 98 | 83.9 | 97.7 | 90.7 | 89.4 | 91.8 | |
| SearchLVLMsbase_model=LLaVA-1.62024.05 | 86.7 | 91.9 | 93.9 | 88.1 | 90.9 | 93.3 | 95.5 | 90.2 | |
| SearchLVLMsbase_model=Gemini 1.5 Pro2024.05 | 82.8 | 79.6 | 91.8 | 81.5 | 97.7 | 84 | 90.9 | 83.3 | |
| SearchLVLMsbase_model=GPT-4V2024.05 | 76.6 | 85.8 | 89.8 | 86.2 | 97.7 | 92 | 87.9 | 87.2 | |
| GPT-4Vvariant=IAG (Information-Augmented Generation)2024.05 | 62.5 | 61.9 | 63.3 | 62.4 | 70.5 | 80 | 69.7 | 64.5 | |
| Gemini 1.5 Provariant=LC (Long Context)2024.05 | 61.7 | 71.5 | 73.5 | 77.2 | 72.7 | 81.3 | 62.1 | 76.4 | |
| GPT-4ovariant=IAG (Information-Augmented Generation)2024.05 | 61.7 | 57.3 | 69.4 | 48.4 | 70.5 | 81.3 | 62.1 | 57.8 | |
| Cham.base_model=InternVL-1.52024.05 | 59.4 | 61.9 | 55.1 | 55.6 | 52.3 | 65.3 | 71.2 | 59.3 | |
| CLIP-FID (C->F)base_model=InternVL-1.5, variant_note=Leverages proposed framework2024.05 | 59.4 | 65 | 44.9 | 54.2 | 47.7 | 65.3 | 53 | 57.7 | |
| Cham.base_model=LLaVA-1.62024.05 | 58.6 | 57.3 | 53.1 | 65.3 | 52.3 | 72 | 74.2 | 62.3 | |
| CLIP-FID (C->F)base_model=LLaVA-1.6, variant_note=Leverages proposed framework2024.05 | 55.5 | 56.5 | 34.7 | 54 | 54.5 | 62.7 | 56.1 | 54.7 | |
| LLaVA-1.6variant=Raw2024.05 | 43.8 | 32.3 | 22.4 | 24.9 | 25 | 53.3 | 33.3 | 31.8 | |
| InternVL-1.5variant=Raw2024.05 | 43.8 | 53.1 | 49 | 29.9 | 34.1 | 73.3 | 37.9 | 42.6 | |
| Gen.base_model=LLaVA-1.62024.05 | 39.1 | 31.5 | 18.4 | 25.7 | 36.4 | 44 | 39.4 | 31.3 | |
| GPT-4ovariant=Raw2024.05 | 36.7 | 34.2 | 42.9 | 28.6 | 40.9 | 65.3 | 36.4 | 37.2 | |
| Gen.base_model=InternVL-1.52024.05 | 29.7 | 28.1 | 28.6 | 22.8 | 31.8 | 42.7 | 28.8 | 27.6 | |
| GPT-4Vvariant=Raw2024.05 | 21.1 | 31.5 | 16.3 | 16.7 | 15.9 | 41.3 | 21.2 | 24.2 | |
| Gemini 1.5 Provariant=Raw2024.05 | 6.2 | 15.8 | 10.2 | 7.4 | 2.3 | 8 | 3 | 9.1 |