CeLLaTe-ner-2class-reinitllrd-tapt-pubmedbert-tokenizer-adapted-combData-gazetteers-lr_3.36e5
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: 0.1191
- Precision: 0.7710
- Recall: 0.7715
- Micro F1: 0.7713
- Weighted F1: 0.7716
- Macro F1: 0.7796
- Accuracy: 0.9840
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: 3.3553919637731326e-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 |
|---|---|---|---|---|---|---|---|---|---|
| 0.1691 | 1.0 | 526 | 0.0682 | 0.5931 | 0.7486 | 0.6619 | 0.6628 | 0.6760 | 0.9754 |
| 0.0159 | 2.0 | 1052 | 0.0743 | 0.7810 | 0.7204 | 0.7495 | 0.7497 | 0.7593 | 0.9834 |
| 0.0078 | 3.0 | 1578 | 0.0803 | 0.7541 | 0.7523 | 0.7532 | 0.7532 | 0.7593 | 0.9834 |
| 0.0042 | 4.0 | 2104 | 0.1010 | 0.6814 | 0.7974 | 0.7348 | 0.7374 | 0.7548 | 0.9805 |
| 0.0029 | 5.0 | 2630 | 0.0922 | 0.7690 | 0.7589 | 0.7639 | 0.7641 | 0.7681 | 0.9844 |
| 0.0023 | 6.0 | 3156 | 0.1128 | 0.7829 | 0.7330 | 0.7571 | 0.7570 | 0.7669 | 0.9833 |
| 0.0017 | 7.0 | 3682 | 0.1158 | 0.8085 | 0.7288 | 0.7666 | 0.7667 | 0.7750 | 0.9836 |
| 0.0013 | 8.0 | 4208 | 0.1334 | 0.8062 | 0.7156 | 0.7582 | 0.7575 | 0.7695 | 0.9825 |
| 0.0012 | 9.0 | 4734 | 0.1144 | 0.7839 | 0.7589 | 0.7712 | 0.7713 | 0.7774 | 0.9839 |
| 0.001 | 10.0 | 5260 | 0.1145 | 0.7751 | 0.7565 | 0.7657 | 0.7657 | 0.7740 | 0.9844 |
| 0.0008 | 11.0 | 5786 | 0.1205 | 0.7857 | 0.7541 | 0.7696 | 0.7696 | 0.7776 | 0.9843 |
| 0.0006 | 12.0 | 6312 | 0.1188 | 0.7706 | 0.7715 | 0.7710 | 0.7714 | 0.7794 | 0.9840 |
| 0.0005 | 13.0 | 6838 | 0.1269 | 0.7808 | 0.7474 | 0.7637 | 0.7638 | 0.7731 | 0.9837 |
| 0.0005 | 14.0 | 7364 | 0.1324 | 0.7712 | 0.7601 | 0.7656 | 0.7659 | 0.7741 | 0.9834 |
| 0.0004 | 15.0 | 7890 | 0.1304 | 0.7889 | 0.7529 | 0.7705 | 0.7705 | 0.7787 | 0.9842 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0