Instructions to use Mardiyyah/no_vague_label_all_false with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Mardiyyah/no_vague_label_all_false with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("token-classification", model="Mardiyyah/no_vague_label_all_false")# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForTokenClassification tokenizer = AutoTokenizer.from_pretrained("Mardiyyah/no_vague_label_all_false") model = AutoModelForTokenClassification.from_pretrained("Mardiyyah/no_vague_label_all_false", device_map="auto") - Notebooks
- Google Colab
- Kaggle
no_vague_label_all_false
This model is a fine-tuned version of microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext on the OTAR3088/CeLLaTe_no_vague1.0_IOB dataset. It achieves the following results on the evaluation set:
- Loss: 0.1240
- Precision: 0.6648
- Recall: 0.7214
- Micro F1: 0.6920
- Weighted F1: 0.6930
- Macro F1: 0.6863
- Accuracy: 0.9798
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 |
|---|---|---|---|---|---|---|---|---|---|
| 0.3409 | 1.0 | 263 | 0.0786 | 0.5949 | 0.6362 | 0.6149 | 0.6146 | 0.6093 | 0.9763 |
| 0.053 | 2.0 | 526 | 0.0829 | 0.5450 | 0.6977 | 0.6119 | 0.6155 | 0.6094 | 0.9737 |
| 0.0362 | 3.0 | 789 | 0.0719 | 0.6047 | 0.6828 | 0.6414 | 0.6450 | 0.6379 | 0.9783 |
| 0.0261 | 4.0 | 1052 | 0.0796 | 0.6211 | 0.6714 | 0.6453 | 0.6471 | 0.6395 | 0.9785 |
| 0.0197 | 5.0 | 1315 | 0.0866 | 0.6346 | 0.7021 | 0.6667 | 0.6690 | 0.6603 | 0.9786 |
| 0.0144 | 6.0 | 1578 | 0.0880 | 0.6486 | 0.6924 | 0.6698 | 0.6701 | 0.6640 | 0.9789 |
| 0.0109 | 7.0 | 1841 | 0.0916 | 0.6769 | 0.6960 | 0.6863 | 0.6866 | 0.6812 | 0.9806 |
| 0.0088 | 8.0 | 2104 | 0.1033 | 0.6633 | 0.7012 | 0.6818 | 0.6837 | 0.6775 | 0.9798 |
| 0.0072 | 9.0 | 2367 | 0.1015 | 0.6598 | 0.6766 | 0.6681 | 0.6673 | 0.6584 | 0.9787 |
| 0.0062 | 10.0 | 2630 | 0.1049 | 0.6724 | 0.6889 | 0.6806 | 0.6806 | 0.6733 | 0.9797 |
| 0.0048 | 11.0 | 2893 | 0.1072 | 0.6596 | 0.7065 | 0.6822 | 0.6836 | 0.6760 | 0.9794 |
| 0.0041 | 12.0 | 3156 | 0.1128 | 0.6751 | 0.7030 | 0.6888 | 0.6891 | 0.6824 | 0.9802 |
| 0.0035 | 13.0 | 3419 | 0.1143 | 0.6515 | 0.7162 | 0.6823 | 0.6839 | 0.6770 | 0.9793 |
| 0.003 | 14.0 | 3682 | 0.1170 | 0.6669 | 0.7162 | 0.6907 | 0.6909 | 0.6837 | 0.9799 |
| 0.0028 | 15.0 | 3945 | 0.1226 | 0.6536 | 0.7162 | 0.6834 | 0.6851 | 0.6784 | 0.9796 |
| 0.0025 | 16.0 | 4208 | 0.1216 | 0.6683 | 0.7153 | 0.6910 | 0.6920 | 0.6857 | 0.9801 |
| 0.0023 | 17.0 | 4471 | 0.1238 | 0.6689 | 0.7100 | 0.6888 | 0.6891 | 0.6828 | 0.9802 |
| 0.002 | 18.0 | 4734 | 0.1244 | 0.6648 | 0.7214 | 0.6920 | 0.6930 | 0.6863 | 0.9798 |
| 0.002 | 19.0 | 4997 | 0.1222 | 0.6711 | 0.7135 | 0.6917 | 0.6924 | 0.6857 | 0.9802 |
| 0.0018 | 20.0 | 5260 | 0.1235 | 0.6645 | 0.7170 | 0.6898 | 0.6907 | 0.6839 | 0.9800 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0
- Downloads last month
- 27