berturk-uncased-keyword-extractor
This model is a fine-tuned version of dbmdz/bert-base-turkish-uncased on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.3931
- Precision: 0.6631
- Recall: 0.6728
- Accuracy: 0.9188
- F1: 0.6679
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
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 8
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | Accuracy | F1 |
|---|---|---|---|---|---|---|---|
| 0.1779 | 1.0 | 1875 | 0.1862 | 0.6199 | 0.6356 | 0.9192 | 0.6276 |
| 0.1327 | 2.0 | 3750 | 0.1890 | 0.6328 | 0.6917 | 0.9206 | 0.6610 |
| 0.1008 | 3.0 | 5625 | 0.2188 | 0.6322 | 0.7037 | 0.9189 | 0.6660 |
| 0.0755 | 4.0 | 7500 | 0.2539 | 0.6395 | 0.7030 | 0.9181 | 0.6697 |
| 0.0574 | 5.0 | 9375 | 0.2882 | 0.6556 | 0.6868 | 0.9197 | 0.6709 |
| 0.0433 | 6.0 | 11250 | 0.3425 | 0.6565 | 0.6851 | 0.9189 | 0.6705 |
| 0.0352 | 7.0 | 13125 | 0.3703 | 0.6616 | 0.6776 | 0.9191 | 0.6695 |
| 0.0288 | 8.0 | 15000 | 0.3931 | 0.6631 | 0.6728 | 0.9188 | 0.6679 |
Framework versions
- Transformers 4.19.2
- Pytorch 1.11.0+cu113
- Datasets 2.2.2
- Tokenizers 0.12.1