KoichiYasuoka's picture
base_model
b6d80ba
|
raw
history blame
960 Bytes
metadata
language:
  - ko
tags:
  - korean
  - masked-lm
base_model: klue/roberta-base
license: cc-by-sa-4.0
pipeline_tag: fill-mask
mask_token: '[MASK]'

roberta-base-korean-hanja

Model Description

This is a RoBERTa model pre-trained on Korean texts, derived from klue/roberta-base. Token-embeddings are enhanced to include all ํ•œ๋ฌธ ๊ต์œก์šฉ ๊ธฐ์ดˆ ํ•œ์ž and ์ธ๋ช…์šฉ ํ•œ์ž characters. You can fine-tune roberta-base-korean-hanja for downstream tasks, such as POS-tagging, dependency-parsing, and so on.

How to Use

from transformers import AutoTokenizer,AutoModelForMaskedLM
tokenizer=AutoTokenizer.from_pretrained("KoichiYasuoka/roberta-base-korean-hanja")
model=AutoModelForMaskedLM.from_pretrained("KoichiYasuoka/roberta-base-korean-hanja")