CeLLaTe-ner-2class-tapt-pubmedbert-tokenizer-adapted-combinedData-lr_3.89
This model is a fine-tuned version of Mardiyyah/CeLLaTe-tapt-pubmedbert-tokenizer-adapted-wwmask-combinedData on the OTAR3088/CeLLaTe-ner-2class-iob_final dataset. It achieves the following results on the evaluation set:
- Loss: 0.1149
- Precision: 0.7696
- Recall: 0.7853
- Micro F1: 0.7774
- Weighted F1: 0.7778
- Macro F1: 0.7857
- Accuracy: 0.9841
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: 3.88548620843886e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 3407
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- 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: linear
- lr_scheduler_warmup_ratio: 0.03
- 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 |
|---|---|---|---|---|---|---|---|---|---|
| 0.1396 | 1.0 | 1052 | 0.0622 | 0.7545 | 0.7372 | 0.7457 | 0.7463 | 0.7586 | 0.9827 |
| 0.0132 | 2.0 | 2104 | 0.0786 | 0.6921 | 0.7679 | 0.7281 | 0.7286 | 0.7434 | 0.9810 |
| 0.0055 | 3.0 | 3156 | 0.0972 | 0.7227 | 0.7709 | 0.7460 | 0.7465 | 0.7590 | 0.9823 |
| 0.0034 | 4.0 | 4208 | 0.1046 | 0.7710 | 0.7673 | 0.7691 | 0.7693 | 0.7763 | 0.9836 |
| 0.0023 | 5.0 | 5260 | 0.1081 | 0.7828 | 0.7432 | 0.7625 | 0.7623 | 0.7710 | 0.9831 |
| 0.0019 | 6.0 | 6312 | 0.1153 | 0.7696 | 0.7853 | 0.7774 | 0.7778 | 0.7857 | 0.9841 |
| 0.0014 | 7.0 | 7364 | 0.1118 | 0.7570 | 0.7980 | 0.7769 | 0.7773 | 0.7846 | 0.9842 |
| 0.0015 | 8.0 | 8416 | 0.1111 | 0.7441 | 0.7799 | 0.7616 | 0.7617 | 0.7664 | 0.9826 |
| 0.0011 | 9.0 | 9468 | 0.1344 | 0.7805 | 0.7571 | 0.7686 | 0.7684 | 0.7740 | 0.9835 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0