berturk-duygu-analizi
This model is a fine-tuned version of dbmdz/bert-base-turkish-cased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2841
- Accuracy: 0.8711
- F1 Macro: 0.8705
- Precision Macro: 0.8718
- Recall Macro: 0.8699
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: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 0.1
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 Macro | Precision Macro | Recall Macro |
|---|---|---|---|---|---|---|---|
| 1.1137 | 0.4673 | 250 | 0.3678 | 0.8031 | 0.8021 | 0.8051 | 0.8050 |
| 0.3016 | 0.9346 | 500 | 0.2808 | 0.8489 | 0.8470 | 0.8463 | 0.8487 |
| 0.2273 | 1.4019 | 750 | 0.2681 | 0.8606 | 0.8604 | 0.8607 | 0.8612 |
| 0.2108 | 1.8692 | 1000 | 0.2583 | 0.8627 | 0.8614 | 0.8623 | 0.8624 |
| 0.1439 | 2.3364 | 1250 | 0.2667 | 0.8688 | 0.8679 | 0.8680 | 0.8702 |
| 0.1511 | 2.8037 | 1500 | 0.2584 | 0.8646 | 0.8637 | 0.8653 | 0.8632 |
| 0.1096 | 3.2710 | 1750 | 0.2676 | 0.8681 | 0.8673 | 0.8675 | 0.8678 |
| 0.1046 | 3.7383 | 2000 | 0.2745 | 0.8690 | 0.8681 | 0.8690 | 0.8681 |
| 0.0823 | 4.2056 | 2250 | 0.2796 | 0.8707 | 0.8703 | 0.8715 | 0.8697 |
| 0.0727 | 4.6729 | 2500 | 0.2841 | 0.8711 | 0.8705 | 0.8718 | 0.8699 |
Framework versions
- Transformers 5.0.0
- Pytorch 2.10.0+cu128
- Datasets 4.8.5
- Tokenizers 0.22.2