BC5CDR_BlueBERT_NER

This model is a fine-tuned version of bionlp/bluebert_pubmed_mimic_uncased_L-12_H-768_A-12 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0944

• Seqeval classification report: precision recall f1-score support

  Chemical 0.84 0.89 0.87 7079 Disease 0.82 0.85 0.83 4968

  micro avg 0.83 0.87 0.85 12047 macro avg 0.83 0.87 0.85 12047

weighted avg 0.83 0.87 0.85 12047

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

Training Loss	Epoch	Step	Validation Loss	Seqeval classification report
No log	1.0	143	0.1111	precision recall f1-score support

Chemical       0.82      0.86      0.84      7079
 Disease       0.76      0.83      0.80      4968

micro avg 0.79 0.85 0.82 12047 macro avg 0.79 0.85 0.82 12047 weighted avg 0.79 0.85 0.82 12047 | | No log | 2.0 | 286 | 0.0987 | precision recall f1-score support

Chemical       0.83      0.89      0.86      7079
 Disease       0.78      0.86      0.82      4968

micro avg 0.81 0.88 0.84 12047 macro avg 0.80 0.87 0.84 12047 weighted avg 0.81 0.88 0.84 12047 | | No log | 3.0 | 429 | 0.0944 | precision recall f1-score support

Chemical       0.84      0.89      0.87      7079
 Disease       0.82      0.85      0.83      4968

micro avg 0.83 0.87 0.85 12047 macro avg 0.83 0.87 0.85 12047 weighted avg 0.83 0.87 0.85 12047 |

Framework versions

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