| ================================================== |
| Training config |
| ================================================== |
| Model name : castorini/afriberta_base |
| Tokenizer name : castorini/afriberta_base |
| Number of labels : 22 |
| Batch size : 8 |
| Learning rate : 5e-05 |
| Number of epochs : 16 |
| Random seed : 42 |
|
|
| ================================================== |
| Epoch 1 |
| ================================================== |
| Traning loss : 818.2853 |
| Validation loss : 0.1867 |
| Precision : 0.8856 |
| Recall : 0.8025 |
| F1-score : 0.8285 |
| Checkpoint saved |
|
|
| ================================================== |
| Epoch 2 |
| ================================================== |
| Traning loss : 245.6919 |
| Validation loss : 0.1653 |
| Precision : 0.9115 |
| Recall : 0.7915 |
| F1-score : 0.8174 |
|
|
| ================================================== |
| Epoch 3 |
| ================================================== |
| Traning loss : 124.5959 |
| Validation loss : 0.1279 |
| Precision : 0.9503 |
| Recall : 0.8683 |
| F1-score : 0.8944 |
| Checkpoint saved |
|
|
| ================================================== |
| Epoch 4 |
| ================================================== |
| Traning loss : 76.5310 |
| Validation loss : 0.2006 |
| Precision : 0.9543 |
| Recall : 0.8714 |
| F1-score : 0.8966 |
| Checkpoint saved |
|
|
| ================================================== |
| Epoch 5 |
| ================================================== |
| Traning loss : 55.0676 |
| Validation loss : 0.1981 |
| Precision : 0.9374 |
| Recall : 0.8800 |
| F1-score : 0.8927 |
|
|
| ================================================== |
| Epoch 6 |
| ================================================== |
| Traning loss : 36.3393 |
| Validation loss : 0.1937 |
| Precision : 0.9037 |
| Recall : 0.8220 |
| F1-score : 0.8429 |
|
|
| ================================================== |
| Epoch 7 |
| ================================================== |
| Traning loss : 21.6568 |
| Validation loss : 0.1819 |
| Precision : 0.8993 |
| Recall : 0.8509 |
| F1-score : 0.8652 |
| Early stopping triggered |
|
|
