Document Visual Question Answering on ArxiVQA
60AccuracyGPT-4o+Ours (Spot-IT)
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4o+Ours (Spot-IT)zero-shot=true2025.08 | 60 | |
| Gem-1.5-Flash+Ours (Spot-IT)zero-shot=true2025.08 | 54 | |
| Gem-1.5-Flashzero-shot=true2025.08 | 53 | |
| GPT-4ozero-shot=true2025.08 | 52 | |
| GPT-4o-mini+Ours (Spot-IT)zero-shot=true2025.08 | 52 | |
| GPT-4o+CoTzero-shot=true2025.08 | 51 | |
| GPT-4o-minizero-shot=true2025.08 | 47 | |
| Qwen2-7Bzero-shot=true2025.08 | 44 | |
| Llama-3.2+Ours (Spot-IT)zero-shot=true2025.08 | 44 | |
| Qwen2-7B+Ours (Spot-IT)zero-shot=true2025.08 | 44 | |
| Llama-3.2+CoTzero-shot=true2025.08 | 42 | |
| GPT-4o+OCRzero-shot=true2025.08 | 41 | |
| Llama-3.2-VL-11Bzero-shot=true2025.08 | 41 | |
| Llama-3.2+OCRzero-shot=true2025.08 | 38 |