newsdata-bertimbal
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: 0.2924
- Accuracy: 0.9183
- Precision: 0.9118
- F1: 0.9144
- Recall: 0.9183
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: 4
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 1
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | F1 | Recall |
|---|---|---|---|---|---|---|---|
| 0.7154 | 0.1024 | 1000 | 0.5830 | 0.856 | 0.8352 | 0.8399 | 0.856 |
| 0.5232 | 0.2048 | 2000 | 0.4769 | 0.874 | 0.8647 | 0.8633 | 0.874 |
| 0.4342 | 0.3071 | 3000 | 0.3966 | 0.891 | 0.8800 | 0.8826 | 0.891 |
| 0.3969 | 0.4095 | 4000 | 0.3509 | 0.9023 | 0.8900 | 0.8949 | 0.9023 |
| 0.3719 | 0.5119 | 5000 | 0.3263 | 0.9102 | 0.9055 | 0.9054 | 0.9102 |
| 0.3638 | 0.6143 | 6000 | 0.3209 | 0.909 | 0.9017 | 0.9035 | 0.909 |
| 0.3217 | 0.7166 | 7000 | 0.3131 | 0.9068 | 0.9025 | 0.9034 | 0.9068 |
| 0.3169 | 0.8190 | 8000 | 0.2952 | 0.9167 | 0.9101 | 0.9125 | 0.9167 |
| 0.3147 | 0.9214 | 9000 | 0.2924 | 0.9183 | 0.9118 | 0.9144 | 0.9183 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.1+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1