metadata
library_name: transformers
license: apache-2.0
language:
- ja
base_model: studio-ousia/luke-japanese-base-lite
pipeline_tag: fill-mask
luke-japanese-base-lite-xlm-roberta
studio-ousia/luke-japanese-base-liteの重みの名前をXLMRoberta形式に置き換え、XLMRobertaモデルとして扱えるようにした物です。
from transformers import XLMRobertaForMaskedLM, XLMRobertaTokenizer
model_name = "hotchpotch/luke-japanese-base-lite-xlm-roberta"
model = XLMRobertaForMaskedLM.from_pretrained(model_name)
tokenizer = XLMRobertaTokenizer.from_pretrained(model_name)