Information Extraction on Real-world documents (test)
94.3Receipts AccuracyClaude-3-5-sonnet-20240620
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Claude-3-5-sonnet-202406202024.10 | 94.3 | 80.5 | 64.6 | 79.8 | |
| H2OVL-Mississippi-2BParams(B)=2.12024.10 | 82 | 56.4 | 41.5 | 59.97 | |
| GPT-4o-202408062024.10 | 80.2 | 81.7 | 77.6 | 79.83 | |
| Intern VL2-40BParams(B)=402024.10 | 80.1 | 68.7 | 56.8 | 68.53 | |
| Intern VL2-26BParams(B)=262024.10 | 76.4 | 64.1 | 43.4 | 61.3 | |
| InternVL2-8BParams(B)=82024.10 | 71.6 | 60.7 | 39.1 | 57.13 | |
| MiniCPM-V-2-6Params(B)=82024.10 | 62 | 58.3 | 46.3 | 55.53 | |
| InternVL2-2BParams(B)=2.12024.10 | 60.5 | 49.2 | 36.5 | 48.73 | |
| InternVL2-4BParams(B)=42024.10 | 58.5 | 53.4 | 42.1 | 51.33 | |
| Intern VL2-1BParams(B)=0.82024.10 | 56.6 | 41.5 | 30.7 | 42.93 |