bert-fraud-classification-test
This model is a fine-tuned version of google-bert/bert-base-multilingual-uncased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.3850
- F1: 0.7886
- Precision: 0.8400
- Val Accuracy: 0.8429
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: 5e-05
- train_batch_size: 44
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 88
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- num_epochs: 2
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | F1 | Precision | Val Accuracy |
|---|---|---|---|---|---|---|
| 0.5784 | 0.1386 | 40 | 0.4995 | 0.6845 | 0.8251 | 0.7873 |
| 0.4545 | 0.2773 | 80 | 0.4295 | 0.7430 | 0.8504 | 0.8200 |
| 0.4566 | 0.4159 | 120 | 0.4116 | 0.7564 | 0.8483 | 0.8266 |
| 0.4468 | 0.5546 | 160 | 0.4149 | 0.7366 | 0.8827 | 0.8217 |
| 0.3454 | 0.6932 | 200 | 0.4062 | 0.7442 | 0.8812 | 0.8254 |
| 0.3333 | 0.8319 | 240 | 0.4046 | 0.7475 | 0.8993 | 0.8296 |
| 0.4618 | 0.9705 | 280 | 0.3973 | 0.7797 | 0.8279 | 0.8358 |
| 0.3581 | 1.1092 | 320 | 0.3869 | 0.7843 | 0.8431 | 0.8410 |
| 0.3946 | 1.2478 | 360 | 0.3869 | 0.7823 | 0.8471 | 0.8405 |
| 0.3263 | 1.3865 | 400 | 0.3875 | 0.7850 | 0.8379 | 0.8405 |
| 0.31 | 1.5251 | 440 | 0.3907 | 0.7721 | 0.8835 | 0.8404 |
| 0.2547 | 1.6638 | 480 | 0.3822 | 0.7855 | 0.8561 | 0.8437 |
| 0.2613 | 1.8024 | 520 | 0.3886 | 0.7883 | 0.8346 | 0.8418 |
| 0.3142 | 1.9411 | 560 | 0.3850 | 0.7886 | 0.8400 | 0.8429 |
Framework versions
- Transformers 4.46.0.dev0
- Pytorch 2.4.0+cu121
- Datasets 3.0.1
- Tokenizers 0.20.0