judithrosell / BC5CDR_ClinicalBERT_NER

BC5CDR_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: 0.1107

• Seqeval classification report: precision recall f1-score support

  Chemical 0.71 0.73 0.72 10493 Disease 0.82 0.82 0.82 6944

  micro avg 0.75 0.77 0.76 17437 macro avg 0.76 0.78 0.77 17437

weighted avg 0.75 0.77 0.76 17437

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

Chemical       0.67      0.68      0.68     10493
 Disease       0.79      0.78      0.78      6944

micro avg 0.72 0.72 0.72 17437 macro avg 0.73 0.73 0.73 17437 weighted avg 0.72 0.72 0.72 17437 | | No log | 2.0 | 286 | 0.1160 | precision recall f1-score support

Chemical       0.69      0.71      0.70     10493
 Disease       0.77      0.83      0.80      6944

micro avg 0.72 0.76 0.74 17437 macro avg 0.73 0.77 0.75 17437 weighted avg 0.72 0.76 0.74 17437 | | No log | 3.0 | 429 | 0.1107 | precision recall f1-score support

Chemical       0.71      0.73      0.72     10493
 Disease       0.82      0.82      0.82      6944

micro avg 0.75 0.77 0.76 17437 macro avg 0.76 0.78 0.77 17437 weighted avg 0.75 0.77 0.76 17437 |

Framework versions

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