<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
frases-bertimbau-v0.4
This model is a fine-tuned version of neuralmind/bert-base-portuguese-cased on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.4380
- F1: 0.8653
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: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 20.0
Training results
| Training Loss | Epoch | Step | Validation Loss | F1 |
|---|---|---|---|---|
| 1.4919 | 0.67 | 500 | 0.6711 | 0.7934 |
| 0.5458 | 1.34 | 1000 | 0.4631 | 0.8445 |
| 0.4346 | 2.01 | 1500 | 0.3971 | 0.8619 |
| 0.3282 | 2.68 | 2000 | 0.3943 | 0.8647 |
| 0.2738 | 3.35 | 2500 | 0.4260 | 0.8651 |
| 0.2261 | 4.02 | 3000 | 0.4071 | 0.8716 |
| 0.1539 | 4.68 | 3500 | 0.4522 | 0.8687 |
| 0.1354 | 5.35 | 4000 | 0.5319 | 0.8633 |
| 0.1132 | 6.02 | 4500 | 0.5306 | 0.8660 |
| 0.0834 | 6.69 | 5000 | 0.5935 | 0.8633 |
| 0.0756 | 7.36 | 5500 | 0.6532 | 0.8593 |
| 0.0692 | 8.03 | 6000 | 0.6492 | 0.8650 |
| 0.0541 | 8.7 | 6500 | 0.6708 | 0.8648 |
| 0.0451 | 9.37 | 7000 | 0.7084 | 0.8667 |
| 0.046 | 10.04 | 7500 | 0.7482 | 0.8655 |
| 0.0356 | 10.71 | 8000 | 0.7802 | 0.8631 |
| 0.0332 | 11.38 | 8500 | 0.8112 | 0.8623 |
| 0.0282 | 12.05 | 9000 | 0.8070 | 0.8664 |
| 0.0251 | 12.72 | 9500 | 0.8332 | 0.8640 |
| 0.0215 | 13.39 | 10000 | 0.8487 | 0.8678 |
| 0.0203 | 14.06 | 10500 | 0.8883 | 0.8604 |
| 0.0168 | 14.73 | 11000 | 0.8870 | 0.8637 |
| 0.0124 | 15.39 | 11500 | 0.8986 | 0.8678 |
| 0.0137 | 16.06 | 12000 | 0.9093 | 0.8670 |
| 0.0104 | 16.73 | 12500 | 0.9145 | 0.8659 |
| 0.0071 | 17.4 | 13000 | 0.9380 | 0.8676 |
| 0.0076 | 18.07 | 13500 | 0.9496 | 0.8686 |
| 0.0072 | 18.74 | 14000 | 0.9589 | 0.8698 |
| 0.0067 | 19.41 | 14500 | 0.9571 | 0.8687 |
Framework versions
- Transformers 4.25.0.dev0
- Pytorch 1.13.0
- Datasets 2.2.1
- Tokenizers 0.13.2
