metadata
base_model: klue/roberta-base
tags:
- generated_from_trainer
- korean
- klue
widget:
- text: 환자는 심부전 진단을 받고 매일 아침 40mg의 푸로세미드를 복용하며, 지속적인 심전도 모니터링을 받습니다.
model-index:
- name: klue-roberta-base-ner-bio
results: []
language:
- ko
pipeline_tag: token-classification
klue-roberta-base-ner-bio
This model is a fine-tuned version of klue/roberta-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0057
- Precision: 0.9888
- Recall: 1.0
- F1: 0.9944
- Accuracy: 0.9998
Model description
간단한 의료 관련 개체명 인식을 제공합니다.
- 약물명 [DR]
- 질병명 [DS]
- 유전자/단백질 명 [GN]
- 임상 증상 [CS]
- 의료 기기 [MD]
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- 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 | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| No log | 1.0 | 30 | 0.0363 | 0.8056 | 0.8898 | 0.8456 | 0.9886 |
| No log | 2.0 | 60 | 0.0079 | 0.9888 | 1.0 | 0.9944 | 0.9998 |
| No log | 3.0 | 90 | 0.0057 | 0.9888 | 1.0 | 0.9944 | 0.9998 |
Framework versions
- Transformers 4.40.2
- Pytorch 2.3.0+cu118
- Datasets 2.19.1
- Tokenizers 0.19.1