<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
BlueBERT
This model is a fine-tuned version of bionlp/bluebert_pubmed_uncased_L-12_H-768_A-12 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.6525
- Accuracy: 0.83
- Precision: 0.8767
- Recall: 0.8889
- F1: 0.8828
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: linear
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 |
|---|---|---|---|---|---|---|---|
| 0.6839 | 1.0 | 50 | 0.7208 | 0.39 | 0.9231 | 0.1667 | 0.2824 |
| 0.6594 | 2.0 | 100 | 0.5862 | 0.6 | 0.9211 | 0.4861 | 0.6364 |
| 0.539 | 3.0 | 150 | 0.5940 | 0.66 | 0.9318 | 0.5694 | 0.7069 |
| 0.4765 | 4.0 | 200 | 0.5675 | 0.65 | 0.9512 | 0.5417 | 0.6903 |
| 0.3805 | 5.0 | 250 | 0.4494 | 0.79 | 0.9322 | 0.7639 | 0.8397 |
| 0.279 | 6.0 | 300 | 0.4760 | 0.84 | 0.8784 | 0.9028 | 0.8904 |
| 0.2016 | 7.0 | 350 | 0.5514 | 0.82 | 0.8553 | 0.9028 | 0.8784 |
| 0.1706 | 8.0 | 400 | 0.5353 | 0.84 | 0.8889 | 0.8889 | 0.8889 |
| 0.1164 | 9.0 | 450 | 0.7676 | 0.82 | 0.8462 | 0.9167 | 0.8800 |
| 0.1054 | 10.0 | 500 | 0.6525 | 0.83 | 0.8767 | 0.8889 | 0.8828 |
Framework versions
- Transformers 4.21.2
- Pytorch 1.12.1+cu113
- Datasets 2.4.0
- Tokenizers 0.12.1
