BERTimbau-base_LeNER-Br
This model is a fine-tuned version of neuralmind/bert-base-portuguese-cased on the lener_br dataset. It achieves the following results on the evaluation set:
- Loss: nan
- Precision: 0.8318
- Recall: 0.8839
- F1: 0.8571
- Accuracy: 0.9754
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| 0.2037 | 1.0 | 979 | nan | 0.7910 | 0.8762 | 0.8314 | 0.9721 |
| 0.0308 | 2.0 | 1958 | nan | 0.7747 | 0.8663 | 0.8180 | 0.9698 |
| 0.02 | 3.0 | 2937 | nan | 0.8316 | 0.8911 | 0.8603 | 0.9801 |
| 0.0133 | 4.0 | 3916 | nan | 0.8038 | 0.8812 | 0.8407 | 0.9728 |
| 0.0111 | 5.0 | 4895 | nan | 0.8253 | 0.8707 | 0.8474 | 0.9753 |
| 0.0078 | 6.0 | 5874 | nan | 0.8235 | 0.8779 | 0.8498 | 0.9711 |
| 0.0057 | 7.0 | 6853 | nan | 0.8174 | 0.8768 | 0.8461 | 0.9760 |
| 0.0032 | 8.0 | 7832 | nan | 0.8113 | 0.8845 | 0.8463 | 0.9769 |
| 0.0027 | 9.0 | 8811 | nan | 0.8344 | 0.8867 | 0.8597 | 0.9767 |
| 0.0023 | 10.0 | 9790 | nan | 0.8318 | 0.8839 | 0.8571 | 0.9754 |
Testing results
metrics={'test_loss': 0.0710107609629631, 'test_precision': 0.8785578747628083, 'test_recall': 0.9138157894736842, 'test_f1': 0.8958400515962593, 'test_accuracy': 0.9884423662270061, 'test_runtime': 12.4395, 'test_samples_per_second': 111.741, 'test_steps_per_second': 13.988})
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1
- Downloads last month
- 1
Finetuned from
Dataset used to train CassioBN/BERTimbau-base_LeNER-Br
Evaluation results
- Precision on lener_brvalidation set self-reported0.832
- Recall on lener_brvalidation set self-reported0.884
- F1 on lener_brvalidation set self-reported0.857
- Accuracy on lener_brvalidation set self-reported0.975