funnel-kor-base / README.md
kykim's picture
Create README.md
b17d58f
|
raw
history blame
450 Bytes
metadata
language: ko

Funnel-transformer 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 FunnelTokenizer, FunnelModel

tokenizer = FunnelTokenizer.from_pretrained("kykim/funnel-kor-base")
model = FunnelModel.from_pretrained("kykim/funnel-kor-base")