| --- |
| license: apache-2.0 |
| datasets: |
| - nguyenthanhasia/vsec-vietnamese-spell-correction |
| - thanhdath/vietnamese-sentences |
| language: |
| - vi |
| metrics: |
| - f1 |
| - precision |
| - recall |
| tags: |
| - spell-correction |
| - vietnamese |
| --- |
| |
| **Dataset**: VSEC | **Records**: 9341 |
|
|
|
|
| ## Metrics tổng hợp |
|
|
| best-327k (30.65M) |
| | Metric | Giá trị | |
| |---|---| |
| | Sentence Exact Match | 0.5693 | |
| | Token Accuracy | 0.9327 | |
| | Char F1 Mean | 0.9924 | |
| | Exact Match trên câu có lỗi | 0.5693 | |
| | Avg latency (ms/câu) | 0.2 | |
| | **Detection Precision** | 0.9568 | |
| | **Detection Recall** | 0.7459 | |
| | **Detection F1** | 0.8383 | |
|
|
| protonx-legal-tc-nano (43.2M) |
| | Metric | Giá trị | |
| |---|---| |
| | Sentence Exact Match | 0.5009 | |
| | Token Accuracy | 0.9401 | |
| | Char F1 Mean | 0.9916 | |
| | Exact Match trên câu có lỗi | 0.5009 | |
| | Avg latency (ms/câu) | 15.8 | |
|
|
| protonx-legal-tc (0.2B) |
| | Metric | Giá trị | |
| |---|---| |
| | Sentence Exact Match | 0.6437 | |
| | Token Accuracy | 0.9489 | |
| | Char F1 Mean | 0.9938 | |
| | Exact Match trên câu có lỗi | 0.6437 | |
| | Avg latency (ms/câu) | 36.9 | |
|
|
|
|
