bert-finetuned-ner
This model is a fine-tuned version of bert-base-cased on the wikiann dataset. It achieves the following results on the evaluation set:
- Loss: 0.3141
- Precision: 0.8169
- Recall: 0.8438
- F1: 0.8301
- Accuracy: 0.9275
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 |
---|---|---|---|---|---|---|---|
0.285 | 1.0 | 2500 | 0.2839 | 0.8005 | 0.8224 | 0.8113 | 0.9201 |
0.2052 | 2.0 | 5000 | 0.2599 | 0.8028 | 0.8383 | 0.8202 | 0.9254 |
0.1419 | 3.0 | 7500 | 0.3141 | 0.8169 | 0.8438 | 0.8301 | 0.9275 |
Framework versions
- Transformers 4.23.0
- Pytorch 1.12.1+cu113
- Datasets 2.5.2
- Tokenizers 0.13.1
- Downloads last month
- 3
Dataset used to train MariaFogh/bert-finetuned-ner
Evaluation results
- Precision on wikiannself-reported0.817
- Recall on wikiannself-reported0.844
- F1 on wikiannself-reported0.830
- Accuracy on wikiannself-reported0.927