judithrosell / JNLPBA_PubMedBERT_NER

JNLPBA_PubMedBERT_NER

This model is a fine-tuned version of microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1450

• Seqeval classification report: precision recall f1-score support

     DNA       0.75      0.83      0.79       955
     RNA       0.80      0.83      0.82      1144
  

  cell_line 0.76 0.79 0.78 5330 cell_type 0.86 0.91 0.88 2518 protein 0.87 0.85 0.86 926

  micro avg 0.80 0.83 0.81 10873 macro avg 0.81 0.84 0.82 10873

weighted avg 0.80 0.83 0.81 10873

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       0.73      0.82      0.77       955
     RNA       0.79      0.82      0.81      1144

cell_line 0.75 0.78 0.76 5330 cell_type 0.86 0.86 0.86 2518 protein 0.86 0.84 0.85 926

micro avg 0.79 0.81 0.80 10873 macro avg 0.80 0.82 0.81 10873 weighted avg 0.79 0.81 0.80 10873 | | 0.145 | 2.0 | 1164 | 0.1473 | precision recall f1-score support

     DNA       0.73      0.82      0.77       955
     RNA       0.85      0.78      0.81      1144

cell_line 0.77 0.78 0.78 5330 cell_type 0.85 0.92 0.88 2518 protein 0.88 0.83 0.85 926

micro avg 0.80 0.82 0.81 10873 macro avg 0.81 0.83 0.82 10873 weighted avg 0.80 0.82 0.81 10873 | | 0.1276 | 3.0 | 1746 | 0.1450 | precision recall f1-score support

     DNA       0.75      0.83      0.79       955
     RNA       0.80      0.83      0.82      1144

cell_line 0.76 0.79 0.78 5330 cell_type 0.86 0.91 0.88 2518 protein 0.87 0.85 0.86 926

micro avg 0.80 0.83 0.81 10873 macro avg 0.81 0.84 0.82 10873 weighted avg 0.80 0.83 0.81 10873 |

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0