OCR-based Visual Question Answering on TextVQA 2019 (val)
83.8AccuracyQwen2-VL
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen2-VL2025.01 | 83.8 | |
| DeepSeekVL-2-Small2025.01 | 83.3 | |
| DeepSeekVL-2-Tiny2025.01 | 80.5 | |
| InternVL2.5-MPO2025.01 | 79.2 | |
| InternVL2.52025.01 | 79 | |
| Qwen2-VL2025.01 | 78.8 | |
| SAIL-VL-8B2025.01 | 77.7 | |
| InternVL22025.01 | 77.6 | |
| InternVL2.5-MPOTraining Strategy=MPO2025.01 | 77.2 | |
| InternVL2.52025.01 | 76.4 | |
| Aquila2025.01 | 76.4 | |
| SAIL-VL2025.01 | 75.7 | |
| InternVL22025.01 | 73.4 |