CeLLaTe-ner-2class-pubmedbert-tapt-tokenizer-adapted-spanmask-gazetters-lr_2.16
This model is a fine-tuned version of Mardiyyah/CeLLaTe-tapt-pubmedbert-tokenizer-adapted-spanmask on the OTAR3088/CeLLaTe-ner-2class-iob_final dataset. It achieves the following results on the evaluation set:
- Loss: 0.0961
- Precision: 0.7974
- Recall: 0.7480
- Micro F1: 0.7720
- Weighted F1: 0.7721
- Macro F1: 0.7820
- Accuracy: 0.9845
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: 2.1601722686e-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: 0.02
- 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 |
|---|---|---|---|---|---|---|---|---|---|
| 0.1486 | 1.0 | 1052 | 0.0586 | 0.7453 | 0.7372 | 0.7412 | 0.7418 | 0.7542 | 0.9827 |
| 0.0156 | 2.0 | 2104 | 0.0823 | 0.7314 | 0.7583 | 0.7446 | 0.7452 | 0.7589 | 0.9812 |
| 0.0066 | 3.0 | 3156 | 0.0874 | 0.7643 | 0.7486 | 0.7564 | 0.7567 | 0.7668 | 0.9825 |
| 0.0038 | 4.0 | 4208 | 0.0961 | 0.7974 | 0.7480 | 0.7720 | 0.7721 | 0.7820 | 0.9845 |
| 0.0026 | 5.0 | 5260 | 0.1134 | 0.6959 | 0.7829 | 0.7368 | 0.7374 | 0.7475 | 0.9806 |
| 0.0022 | 6.0 | 6312 | 0.1119 | 0.7533 | 0.7601 | 0.7567 | 0.7567 | 0.7583 | 0.9833 |
| 0.0015 | 7.0 | 7364 | 0.1115 | 0.7887 | 0.7589 | 0.7735 | 0.7735 | 0.7801 | 0.9838 |
| 0.0014 | 8.0 | 8416 | 0.1053 | 0.7406 | 0.7673 | 0.7537 | 0.7540 | 0.7626 | 0.9830 |
| 0.001 | 9.0 | 9468 | 0.1208 | 0.7591 | 0.7673 | 0.7632 | 0.7635 | 0.7730 | 0.9835 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0