LAM dataset results
| Model | CER | WER | WER BoW |
|---|---|---|---|
| Gpt-4o-mini-2024-07-18 | 37.92% | 52.98% | 44.77% |
| Gpt-4o-2024-08-06 | 36.56% | 52.32% | 46.71% |
| Claude-3-5-sonnet-20240620 | 20.08% | 34.43% | 24.56% |
| MinicpmV-2 6 | 43.62% | 68.84% | 61.59% |
| Qwen2-VL-7B | 31.11% | 52.11% | 44.25% |
| Pixtral-12B | 54.17% | 79.25% | 64.69% |
| InternVL2-8B | 80.04% | 98.41% | 92.81% |
| Phi-3-mini-128k-instruct | 73.77% | 93.18% | 90.09% |
| Model | CER | WER | WER BoW |
|---|---|---|---|
| Gpt-4o-mini-2024-07-18 | 37.92% | 52.98% | 44.77% |
| Gpt-4o-2024-08-06 | 36.56% | 52.32% | 46.71% |
| Claude-3-5-sonnet-20240620 | |||
| MinicpmV-2 6 | 43.62% | 68.84% | 61.59% |
| Qwen2-VL-7B | 31.11% | 52.11% | 44.25% |
| Pixtral-12B | 54.17% | 79.25% | 64.69% |
| InternVL2-8B | 80.04% | 98.41% | 92.81% |
| Phi-3-mini-128k-instruct | 73.77% | 93.18% | 90.09% |