您好，我下载代码后，CDR数据集的loss稳定在0.693后就不再下降，acc也只有0.5左右，请问是什么原因？

chentao1999 / MedicalRelationExtraction

The depository support training and testing BERT-CNN model on three medical relation extraction corpora: BioCreative V CDR task corpus, traditional Chinese medicine literature corpus, and i2b2 temporal relation corpus.

MIT License

72 stars 20 forks source link

您好，我下载代码后，CDR数据集的loss稳定在0.693后就不再下降，acc也只有0.5左右，请问是什么原因？ #4

Open feelingnotes opened 4 years ago

feelingnotes commented 4 years ago

Hi, after I download and run the code, the minimal loss of CDR dataset is around 0.693, the accuracy is also around 0.5, do you know why?

zelin-x commented 2 years ago

hi, I meet the same problem, have you already solved it? thanks