Leopardi dataset results
| Model | CER | WER | WER BoW |
|---|---|---|---|
| Italian Handwriting M1 | 37.23% | 55.77% | 47.34% |
| Gpt-4o-mini-2024-07-18 | 48.13% | 67.82% | 55.46% |
| Gpt-4o-2024-08-06 | 36.34% | 51.67% | 43.82% |
| Claude-3-5-sonnet-20240620 | 26.43% | 35.08% | 31.15% |
| MinicpmV-2 6 | 45.85% | 65.73% | 60.82% |
| Qwen2-VL-7B | 37.07% | 49.07% | 43.91% |
| Pixtral-12B | 59.54% | 82.02% | 68.50% |
| InternVL2-8B | 83.43% | 99.70% | 97.11% |
| Phi-3-mini-128k-instruct | 69.78% | 85.06% | 82.01% |
| Model | CER | WER | WER BoW |
|---|---|---|---|
| Italian Handwriting M1 | 37.23% | 55.77% | 47.34% |
| Gpt-4o-mini-2024-07-18 | 48.13% | 67.82% | 55.46% |
| Gpt-4o-2024-08-06 | 36.34% | 51.67% | 43.82% |
| Claude-3-5-sonnet-20240620 | |||
| MinicpmV-2 6 | 45.85% | 65.73% | 60.82% |
| Qwen2-VL-7B | 37.07% | 49.07% | 43.91% |
| Pixtral-12B | 59.54% | 82.02% | 68.50% |
| InternVL2-8B | 83.43% | 99.70% | 97.11% |
| Phi-3-mini-128k-instruct | 69.78% | 85.06% | 82.01% |