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")