bert-finetuned-ner
This model is a fine-tuned version of bert-base-cased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 2.5658
- Precision: 1.0
- Recall: 1.0
- F1: 1.0
- Accuracy: 1.0
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: 8
- eval_batch_size: 8
- seed: 42
- 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 |
|---|---|---|---|---|---|---|---|
| No log | 1.0 | 1 | 2.8381 | 1.0 | 1.0 | 1.0 | 1.0 |
| No log | 2.0 | 2 | 2.6555 | 1.0 | 1.0 | 1.0 | 1.0 |
| No log | 3.0 | 3 | 2.5658 | 1.0 | 1.0 | 1.0 | 1.0 |
Framework versions
- Transformers 4.27.4
- Pytorch 2.0.0+cu118
- Datasets 2.11.0
- Tokenizers 0.13.3