finetuned_parsBERT_NER_fa
This model is a fine-tuned version of HooshvareLab/bert-fa-zwnj-base on the mixed NER dataset collected from ARMAN, PEYMA, and WikiANN. It achieves the following results on the evaluation set:
- Loss: 0.0297
- Precision: 0.9481
- Recall: 0.9582
- F1: 0.9531
- Accuracy: 0.9942
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: 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: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| 0.12 | 1.0 | 1821 | 0.0543 | 0.8387 | 0.8577 | 0.8481 | 0.9830 |
| 0.0381 | 2.0 | 3642 | 0.0360 | 0.8941 | 0.9247 | 0.9091 | 0.9898 |
| 0.0168 | 3.0 | 5463 | 0.0282 | 0.9273 | 0.9452 | 0.9362 | 0.9927 |
| 0.0078 | 4.0 | 7284 | 0.0284 | 0.9391 | 0.9551 | 0.9470 | 0.9938 |
| 0.0033 | 5.0 | 9105 | 0.0297 | 0.9481 | 0.9582 | 0.9531 | 0.9942 |
Framework versions
- Transformers 4.21.2
- Pytorch 1.12.1+cu113
- Datasets 2.4.0
- Tokenizers 0.12.1