CeLLaTe-ner-2class-pubmedbert-tapt-combData-tokenizer-original-wwmask
This model is a fine-tuned version of Mardiyyah/CeLLaTe-tapt-pubmedbert-tokenizer-original-wwmask-combinedData on the OTAR3088/CeLLaTe-ner-2class-iob_final dataset. It achieves the following results on the evaluation set:
- Loss: 0.0575
- Precision: 0.7786
- Recall: 0.7425
- Micro F1: 0.7601
- Weighted F1: 0.7604
- Macro F1: 0.7695
- Accuracy: 0.9837
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 | Micro F1 | Weighted F1 | Macro F1 | Accuracy |
|---|---|---|---|---|---|---|---|---|---|
| 0.2519 | 1.0 | 263 | 0.0564 | 0.6725 | 0.7328 | 0.7013 | 0.7036 | 0.7220 | 0.9819 |
| 0.0404 | 2.0 | 526 | 0.0539 | 0.7758 | 0.7200 | 0.7468 | 0.7471 | 0.7581 | 0.9835 |
| 0.0266 | 3.0 | 789 | 0.0572 | 0.7786 | 0.7425 | 0.7601 | 0.7604 | 0.7695 | 0.9837 |
| 0.0196 | 4.0 | 1052 | 0.0613 | 0.7573 | 0.7462 | 0.7517 | 0.7521 | 0.7605 | 0.9841 |
| 0.0147 | 5.0 | 1315 | 0.0699 | 0.7933 | 0.7261 | 0.7582 | 0.7582 | 0.7621 | 0.9838 |
| 0.0119 | 6.0 | 1578 | 0.0740 | 0.7395 | 0.6980 | 0.7181 | 0.7181 | 0.7158 | 0.9825 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0