| 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](https://github.com/kiyoungkim1/LM-kor) | |
| ```python | |
| from transformers import FunnelTokenizer, FunnelModel | |
| tokenizer = FunnelTokenizer.from_pretrained("kykim/funnel-kor-base") | |
| model = FunnelModel.from_pretrained("kykim/funnel-kor-base") | |
| ``` |