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")