CeLLaTe-ner-2class-bioformer16l-reinitllrd-with-gazetteers-lr_2754e-5
This model is a fine-tuned version of bioformers/bioformer-16l on the OTAR3088/CeLLaTe-ner-2class-iob_final dataset. It achieves the following results on the evaluation set:
- Loss: 0.0791
- Precision: 0.8043
- Recall: 0.7683
- Micro F1: 0.7859
- Weighted F1: 0.7857
- Macro F1: 0.7946
- Accuracy: 0.9834
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: 2.754605200107782e-05
- train_batch_size: 32
- eval_batch_size: 16
- seed: 17
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.01
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | Micro F1 | Weighted F1 | Macro F1 | Accuracy |
|---|---|---|---|---|---|---|---|---|---|
| 0.1589 | 1.0 | 526 | 0.0656 | 0.7103 | 0.7008 | 0.7055 | 0.7071 | 0.7190 | 0.9781 |
| 0.0225 | 2.0 | 1052 | 0.0808 | 0.6590 | 0.6878 | 0.6731 | 0.6761 | 0.6896 | 0.9761 |
| 0.0121 | 3.0 | 1578 | 0.0788 | 0.8043 | 0.7683 | 0.7859 | 0.7857 | 0.7946 | 0.9834 |
| 0.0073 | 4.0 | 2104 | 0.0881 | 0.7331 | 0.7567 | 0.7447 | 0.7452 | 0.7586 | 0.9796 |
| 0.005 | 5.0 | 2630 | 0.1014 | 0.6844 | 0.7352 | 0.7089 | 0.7095 | 0.7173 | 0.9775 |
| 0.0033 | 6.0 | 3156 | 0.1012 | 0.7805 | 0.7791 | 0.7798 | 0.7797 | 0.7888 | 0.9824 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0