distilbert-base-uncased_ner_conll2003
This model is a fine-tuned version of distilbert-base-uncased on the conll2003 dataset. It achieves the following results on the evaluation set:
- Loss: 0.0524
- Precision: 0.9358
- Recall: 0.9438
- F1: 0.9398
- Accuracy: 0.9877
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: cosine
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| 0.1897 | 1.0 | 878 | 0.0544 | 0.9223 | 0.9270 | 0.9246 | 0.9848 |
| 0.0363 | 2.0 | 1756 | 0.0486 | 0.9316 | 0.9391 | 0.9353 | 0.9869 |
| 0.0194 | 3.0 | 2634 | 0.0496 | 0.9369 | 0.9403 | 0.9386 | 0.9873 |
| 0.0114 | 4.0 | 3512 | 0.0526 | 0.9340 | 0.9436 | 0.9388 | 0.9875 |
| 0.0089 | 5.0 | 4390 | 0.0524 | 0.9358 | 0.9438 | 0.9398 | 0.9877 |
Framework versions
- Transformers 4.20.1
- Pytorch 1.11.0
- Datasets 2.1.0
- Tokenizers 0.12.1