Fine-tuned BERTImbau for legal texts classification
This model is a fine-tuned version of neuralmind/bert-large-portuguese-cased on a dataset containing summaries of TJSP decisions, with the purpose of classyfing the text on 5 legal areas. It achieves the following results on the evaluation set:
- Loss: 0.5813
- Accuracy: 0.8713
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-06
- 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: 3.0
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.2709 | 1.0 | 8509 | 0.5307 | 0.8388 |
| 0.2388 | 2.0 | 17018 | 0.4947 | 0.8692 |
| 0.1761 | 3.0 | 25527 | 0.5813 | 0.8713 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.1+cu118
- Datasets 2.20.0
- Tokenizers 0.19.1
- Downloads last month
- 3