CeLLaTe-ner-bioformer16l-tapt-tokenizer-adapted-v2
This model is a fine-tuned version of Mardiyyah/CeLLaTe-tapt-bioformer16l-tokenizer-adapted_v2 on the Mardiyyah/CeLLaTe-ner-2class-iob dataset. It achieves the following results on the evaluation set:
- Loss: 0.0939
- Precision: 0.7488
- Recall: 0.7695
- F1: 0.7590
- Accuracy: 0.9821
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: 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: 0.01
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| 0.2372 | 1.0 | 263 | 0.0680 | 0.5850 | 0.6761 | 0.6273 | 0.9771 |
| 0.0422 | 2.0 | 526 | 0.0577 | 0.7021 | 0.7220 | 0.7119 | 0.9811 |
| 0.0298 | 3.0 | 789 | 0.0580 | 0.6936 | 0.7537 | 0.7224 | 0.9810 |
| 0.0225 | 4.0 | 1052 | 0.0621 | 0.7113 | 0.7005 | 0.7058 | 0.9807 |
| 0.0179 | 5.0 | 1315 | 0.0661 | 0.7096 | 0.7180 | 0.7138 | 0.9808 |
| 0.0146 | 6.0 | 1578 | 0.0716 | 0.7262 | 0.6908 | 0.7081 | 0.9804 |
| 0.0115 | 7.0 | 1841 | 0.0721 | 0.7441 | 0.7458 | 0.7449 | 0.9819 |
| 0.01 | 8.0 | 2104 | 0.0804 | 0.7364 | 0.7180 | 0.7271 | 0.9812 |
| 0.008 | 9.0 | 2367 | 0.0853 | 0.74 | 0.7123 | 0.7259 | 0.9817 |
| 0.0075 | 10.0 | 2630 | 0.0848 | 0.7602 | 0.7486 | 0.7544 | 0.9825 |
| 0.0063 | 11.0 | 2893 | 0.0881 | 0.7521 | 0.7458 | 0.7489 | 0.9821 |
| 0.0057 | 12.0 | 3156 | 0.0910 | 0.7667 | 0.7333 | 0.7496 | 0.9823 |
| 0.005 | 13.0 | 3419 | 0.0945 | 0.7488 | 0.7695 | 0.7590 | 0.9821 |
| 0.0045 | 14.0 | 3682 | 0.0963 | 0.7680 | 0.7424 | 0.7550 | 0.9825 |
| 0.004 | 15.0 | 3945 | 0.0945 | 0.7426 | 0.7627 | 0.7525 | 0.9819 |
| 0.0038 | 16.0 | 4208 | 0.0982 | 0.7625 | 0.7508 | 0.7566 | 0.9822 |
| 0.0035 | 17.0 | 4471 | 0.1045 | 0.7816 | 0.7135 | 0.7460 | 0.9820 |
| 0.0032 | 18.0 | 4734 | 0.1013 | 0.7474 | 0.7588 | 0.7530 | 0.9823 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0