CeLLaTe-ner-2class-bioformer16l-baseline-with-gazetteers
This model is a fine-tuned version of bioformers/bioformer-16l on the OTAR3088/CeLLaTe-ner-2class-iob_final dataset. It achieves the following results on the evaluation set:
- Loss: 0.1372
- Precision: 0.8086
- Recall: 0.7764
- Micro F1: 0.7922
- Weighted F1: 0.7921
- Macro F1: 0.8023
- Accuracy: 0.9830
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 |
|---|---|---|---|---|---|---|---|---|---|
| 0.1326 | 1.0 | 1052 | 0.0608 | 0.7428 | 0.7464 | 0.7446 | 0.7455 | 0.7580 | 0.9811 |
| 0.0196 | 2.0 | 2104 | 0.0765 | 0.7646 | 0.7308 | 0.7473 | 0.7479 | 0.7566 | 0.9805 |
| 0.0096 | 3.0 | 3156 | 0.0927 | 0.7879 | 0.7545 | 0.7708 | 0.7707 | 0.7811 | 0.9820 |
| 0.0062 | 4.0 | 4208 | 0.0968 | 0.7803 | 0.7464 | 0.7630 | 0.7634 | 0.7751 | 0.9808 |
| 0.004 | 5.0 | 5260 | 0.1049 | 0.7920 | 0.7424 | 0.7664 | 0.7663 | 0.7717 | 0.9816 |
| 0.0027 | 6.0 | 6312 | 0.1140 | 0.7596 | 0.7419 | 0.7507 | 0.7511 | 0.7603 | 0.9809 |
| 0.0021 | 7.0 | 7364 | 0.1261 | 0.7399 | 0.7303 | 0.7351 | 0.7360 | 0.7437 | 0.9798 |
| 0.0017 | 8.0 | 8416 | 0.1148 | 0.7677 | 0.7818 | 0.7747 | 0.7753 | 0.7891 | 0.9819 |
| 0.0014 | 9.0 | 9468 | 0.1204 | 0.7900 | 0.7670 | 0.7783 | 0.7785 | 0.7899 | 0.9826 |
| 0.0013 | 10.0 | 10520 | 0.1234 | 0.7933 | 0.7809 | 0.7870 | 0.7874 | 0.7980 | 0.9832 |
| 0.001 | 11.0 | 11572 | 0.1336 | 0.7797 | 0.7835 | 0.7816 | 0.7820 | 0.7941 | 0.9822 |
| 0.0009 | 12.0 | 12624 | 0.1339 | 0.7959 | 0.7710 | 0.7833 | 0.7832 | 0.7951 | 0.9824 |
| 0.0008 | 13.0 | 13676 | 0.1308 | 0.7982 | 0.7818 | 0.7899 | 0.7900 | 0.8008 | 0.9831 |
| 0.0007 | 14.0 | 14728 | 0.1363 | 0.8087 | 0.7639 | 0.7856 | 0.7856 | 0.7941 | 0.9827 |
| 0.0006 | 15.0 | 15780 | 0.1374 | 0.8086 | 0.7764 | 0.7922 | 0.7921 | 0.8023 | 0.9830 |
| 0.0005 | 16.0 | 16832 | 0.1387 | 0.8021 | 0.7688 | 0.7851 | 0.7854 | 0.7962 | 0.9830 |
| 0.0005 | 17.0 | 17884 | 0.1423 | 0.7983 | 0.7737 | 0.7858 | 0.7860 | 0.7968 | 0.9829 |
| 0.0005 | 18.0 | 18936 | 0.1443 | 0.7982 | 0.7750 | 0.7865 | 0.7866 | 0.7968 | 0.9829 |
| 0.0004 | 19.0 | 19988 | 0.1471 | 0.8068 | 0.7733 | 0.7897 | 0.7896 | 0.7998 | 0.9828 |
| 0.0004 | 19.9815 | 21020 | 0.1459 | 0.7976 | 0.7773 | 0.7873 | 0.7874 | 0.7976 | 0.9827 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0