LAM dataset results
| Model | CER | WER | WER BoW |
|---|---|---|---|
| Italian Handwriting M1 | 25.94% | 45.77% | 36.75% |
| Gpt-4o-mini-2024-07-18 | 37.86% | 60.83% | 52.25% |
| Gpt-4o-2024-08-06 | 33.53% | 49.81% | 42.53% |
| Claude-3-5-sonnet-20240620 | 20.55% | 27.78% | 24.57% |
| MinicpmV-2 6 | 41.55% | 62.56% | 61.05% |
| Qwen2-VL-7B | 28.76% | 42.46% | 40.80% |
| Pixtral-12B | 52.73% | 64.40% | 62.44% |
| InternVL2-8B | 80.82% | 94.22% | 93.24% |
| Phi-3-mini-128k-instruct | 64.78% | 81.42% | 80.12% |
| Model | CER | WER | WER BoW |
|---|---|---|---|
| Italian Handwriting M1 | 25.94% | 45.77% | 36.75% |
| Gpt-4o-mini-2024-07-18 | 37.86% | 60.83% | 52.25% |
| Gpt-4o-2024-08-06 | 33.53% | 49.81% | 42.53% |
| Claude-3-5-sonnet-20240620 | |||
| MinicpmV-2 6 | 41.55% | 62.56% | 61.05% |
| Qwen2-VL-7B | 28.76% | 42.46% | 40.80% |
| Pixtral-12B | 52.73% | 64.40% | 62.44% |
| InternVL2-8B | 80.82% | 94.22% | 93.24% |
| Phi-3-mini-128k-instruct | 64.78% | 81.42% | 80.12% |