Document and chart understanding on AI2D (Pass@1)
88.7Pass@1Qwen 2.5-VL 72B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen 2.5-VL 72Bthinking=false, decoding=greedy2025.05 | 88.7 | |
| Seed 1.5-VLthinking=false, decoding=greedy2025.05 | 88.5 | |
| Gemini 1.5 Prothinking=true, decoding=greedy2025.05 | 88.4 | |
| Seed 1.5-VLthinking=true, decoding=greedy2025.05 | 87.3 | |
| GPT-4othinking=false, decoding=greedy2025.05 | 84.9 | |
| Claude 3.7 Sonnetthinking=true, decoding=sampling2025.05 | 82.1 | |
| OpenAI o1thinking=true, decoding=greedy2025.05 | 79.5 |