bert-large-uncased-en-ner
This model is a fine-tuned version of bert-large-uncased on the conll2003 dataset. It achieves the following results on the evaluation set:
- Loss: 0.1402
- Precision: 0.9094
- Recall: 0.9164
- F1: 0.9129
- Accuracy: 0.9791
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
The model was trained on data that follows the IOB convention. Full tagset with indices:
{'O': 0, 'B-PER': 1, 'I-PER': 2, 'B-ORG': 3, 'I-ORG': 4, 'B-LOC': 5, 'I-LOC': 6, 'B-MISC': 7, 'I-MISC': 8}
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 4e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 0
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| 0.0787 | 1.0 | 1756 | 0.1122 | 0.8932 | 0.9046 | 0.8989 | 0.9761 |
| 0.0406 | 2.0 | 3512 | 0.1312 | 0.9042 | 0.9124 | 0.9083 | 0.9777 |
| 0.0177 | 3.0 | 5268 | 0.1402 | 0.9094 | 0.9164 | 0.9129 | 0.9791 |
Framework versions
- Transformers 4.27.2
- Pytorch 2.0.0+cu117
- Datasets 2.10.1
- Tokenizers 0.13.2