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: PubMedBERT-full-finetuned-ner-pablo
    results: []

PubMedBERT-full-finetuned-ner-pablo

This model is a fine-tuned version of microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext on the n2c2 2018 dataset for the paper https://arxiv.org/abs/2409.19467. It achieves the following results on the evaluation set:

Loss: 0.0712
Precision: 0.8087
Recall: 0.7954
F1: 0.8020
Accuracy: 0.9781

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: 5e-05
train_batch_size: 32
eval_batch_size: 32
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 4
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	231	0.0934	0.7464	0.7652	0.7557	0.9730
No log	2.0	462	0.0730	0.7975	0.7915	0.7945	0.9774
0.2789	3.0	693	0.0713	0.8075	0.7924	0.7999	0.9777
0.2789	4.0	924	0.0712	0.8087	0.7954	0.8020	0.9781

Framework versions

Transformers 4.44.2
Pytorch 2.4.1+cu121
Datasets 2.21.0
Tokenizers 0.19.1