berturk-keyword-extractor
This model is a fine-tuned version of dbmdz/bert-base-turkish-cased on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.4306
- Precision: 0.6770
- Recall: 0.6899
- Accuracy: 0.9169
- F1: 0.6834
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 |
|---|---|---|---|---|---|---|---|
| 0.1845 | 1.0 | 1875 | 0.1964 | 0.6380 | 0.6743 | 0.9164 | 0.6557 |
| 0.1338 | 2.0 | 3750 | 0.2023 | 0.6407 | 0.7081 | 0.9169 | 0.6727 |
| 0.0978 | 3.0 | 5625 | 0.2315 | 0.6434 | 0.7309 | 0.9159 | 0.6844 |
| 0.0742 | 4.0 | 7500 | 0.2746 | 0.6592 | 0.7144 | 0.9158 | 0.6857 |
| 0.0541 | 5.0 | 9375 | 0.3290 | 0.6700 | 0.6880 | 0.9161 | 0.6789 |
| 0.0426 | 6.0 | 11250 | 0.3608 | 0.6789 | 0.6860 | 0.9171 | 0.6824 |
| 0.0332 | 7.0 | 13125 | 0.4075 | 0.6769 | 0.6924 | 0.9168 | 0.6845 |
| 0.027 | 8.0 | 15000 | 0.4306 | 0.6770 | 0.6899 | 0.9169 | 0.6834 |
Framework versions
- Transformers 4.19.2
- Pytorch 1.11.0+cu113
- Datasets 2.2.2
- Tokenizers 0.12.1