CeLLaTe-ner-3class-bioformer8l-baseline
This model is a fine-tuned version of bioformers/bioformer-8l on the OTAR3088/CeLLaTe-ner-3class-iob_final dataset. It achieves the following results on the evaluation set:
- Loss: 0.0904
- Precision: 0.7620
- Recall: 0.7251
- Micro F1: 0.7431
- Weighted F1: 0.7414
- Macro F1: 0.7292
- Accuracy: 0.9796
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 |
|---|---|---|---|---|---|---|---|---|---|
| 0.368 | 1.0 | 263 | 0.1009 | 0.4522 | 0.4864 | 0.4687 | 0.4498 | 0.4181 | 0.9666 |
| 0.0766 | 2.0 | 526 | 0.0744 | 0.6974 | 0.6853 | 0.6913 | 0.6926 | 0.6786 | 0.9761 |
| 0.0477 | 3.0 | 789 | 0.0741 | 0.7137 | 0.7148 | 0.7143 | 0.7116 | 0.7009 | 0.9776 |
| 0.0353 | 4.0 | 1052 | 0.0792 | 0.7014 | 0.6475 | 0.6734 | 0.6726 | 0.6556 | 0.9762 |
| 0.0269 | 5.0 | 1315 | 0.0818 | 0.6886 | 0.7097 | 0.6990 | 0.6994 | 0.6930 | 0.9767 |
| 0.0211 | 6.0 | 1578 | 0.0860 | 0.6955 | 0.6837 | 0.6896 | 0.6905 | 0.6836 | 0.9768 |
| 0.0177 | 7.0 | 1841 | 0.0887 | 0.7186 | 0.7271 | 0.7228 | 0.7230 | 0.7147 | 0.9777 |
| 0.0149 | 8.0 | 2104 | 0.0911 | 0.7620 | 0.7251 | 0.7431 | 0.7414 | 0.7292 | 0.9796 |
| 0.0121 | 9.0 | 2367 | 0.0890 | 0.7216 | 0.7515 | 0.7363 | 0.7352 | 0.7230 | 0.9785 |
| 0.0103 | 10.0 | 2630 | 0.1008 | 0.7319 | 0.7334 | 0.7326 | 0.7311 | 0.7183 | 0.9782 |
| 0.0091 | 11.0 | 2893 | 0.1026 | 0.7439 | 0.7184 | 0.7309 | 0.7297 | 0.7158 | 0.9789 |
| 0.0086 | 12.0 | 3156 | 0.1044 | 0.7168 | 0.6999 | 0.7083 | 0.7089 | 0.6957 | 0.9780 |
| 0.0072 | 13.0 | 3419 | 0.1105 | 0.7286 | 0.7148 | 0.7217 | 0.7215 | 0.7092 | 0.9781 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0