benchaffe's picture
Update README.md
fde8a9e verified
metadata
library_name: transformers
license: mit
base_model: microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext
tags:
  - generated_from_trainer
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: BiomedBERT-AC-LF-Classification
    results: []
datasets:
  - surrey-nlp/PLOD-CW-25
language:
  - en

BiomedBERT-AC-LF-Classification

This model is a fine-tuned version of microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext on the PLOD-CW-25 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2703
  • Precision: 0.7821
  • Recall: 0.8686
  • F1: 0.8231
  • Accuracy: 0.9204

It achieves the following results on the test set:

  • Loss: 0.1384
  • Precision: 0.8473
  • Recall: 0.9281
  • F1: 0.8858
  • Accuracy: 0.9529

Model description

This model is a fine-tuned model, designed to detect abbreviations and long forms in biomedical text. Abbreviations and long forms are tagged in the BIO format, with the following labels, B-AC, B-LF, I-LF and O.

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
  • 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
  • num_epochs: 5

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
0.3341 1.0 125 0.2485 0.7727 0.8477 0.8084 0.9111
0.1633 2.0 250 0.2525 0.7767 0.8673 0.8195 0.9174
0.1293 3.0 375 0.2224 0.7855 0.8501 0.8165 0.9211
0.1081 4.0 500 0.2600 0.7780 0.8784 0.8252 0.9201
0.0938 5.0 625 0.2703 0.7821 0.8686 0.8231 0.9204

Framework versions

  • Transformers 4.52.4
  • Pytorch 2.6.0+cu124
  • Datasets 3.6.0
  • Tokenizers 0.21.1