README.md

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")