albert-kor-base / README.md
kykim's picture
Create README.md
04e79bc
metadata
language: ko

Albert base model for Korean

  • 70GB Korean text dataset and 42000 lower-cased subwords are used
  • Check the model performance and other language models for Korean in github
from transformers import BertTokenizerFast, AlbertModel

tokenizer_albert = BertTokenizerFast.from_pretrained("kykim/albert-kor-base")
model_albert = AlbertModel.from_pretrained("kykim/albert-kor-base")