judithrosell / JNLPBA_ClinicalBERT_NER

JNLPBA_ClinicalBERT_NER

This model is a fine-tuned version of medicalai/ClinicalBERT on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1723

• Seqeval classification report: precision recall f1-score support

     DNA       0.72      0.81      0.77      1351
     RNA       0.71      0.86      0.78       723
  

  cell_line 0.84 0.74 0.78 582 cell_type 0.72 0.75 0.73 5623 protein 0.85 0.85 0.85 3501

  micro avg 0.76 0.79 0.78 11780 macro avg 0.77 0.80 0.78 11780

weighted avg 0.76 0.79 0.78 11780

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: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

     DNA       0.72      0.77      0.75      1351
     RNA       0.70      0.84      0.77       723

cell_line 0.85 0.70 0.77 582 cell_type 0.71 0.68 0.69 5623 protein 0.85 0.80 0.83 3501

micro avg 0.76 0.74 0.75 11780 macro avg 0.77 0.76 0.76 11780 weighted avg 0.76 0.74 0.75 11780 | | 0.1841 | 2.0 | 1164 | 0.1762 | precision recall f1-score support

     DNA       0.73      0.78      0.76      1351
     RNA       0.70      0.87      0.78       723

cell_line 0.86 0.71 0.78 582 cell_type 0.71 0.73 0.72 5623 protein 0.86 0.83 0.84 3501

micro avg 0.76 0.77 0.77 11780 macro avg 0.77 0.78 0.78 11780 weighted avg 0.77 0.77 0.77 11780 | | 0.1582 | 3.0 | 1746 | 0.1723 | precision recall f1-score support

     DNA       0.72      0.81      0.77      1351
     RNA       0.71      0.86      0.78       723

cell_line 0.84 0.74 0.78 582 cell_type 0.72 0.75 0.73 5623 protein 0.85 0.85 0.85 3501

micro avg 0.76 0.79 0.78 11780 macro avg 0.77 0.80 0.78 11780 weighted avg 0.76 0.79 0.78 11780 |

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0