OCR Webpage Document on Text-rich VQA
83.7AccuracyLLaVA-OneVision-72B
Evaluation results
| Method | Links | |
|---|---|---|
| LLaVA-OneVision-72BModel Scale=72B2024.08 | 83.7 | |
| LLaVA-OneVision-7BModel Scale=7B2024.08 | 80.1 | |
| LLaVA-OneVision-0.5BModel Scale=0.5B2024.08 | 65 | |
| GPT-4VVersion=V-Preview2024.08 | 54.5 |