deberta-base-japanese-wikipedia

Model Description

This is a DeBERTa(V2) model pre-trained on Japanese Wikipedia and 青空文庫 texts. NVIDIA A100-SXM4-40GB took 109 hours 27 minutes for training. You can fine-tune deberta-base-japanese-wikipedia 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/deberta-base-japanese-wikipedia")
model=AutoModelForMaskedLM.from_pretrained("KoichiYasuoka/deberta-base-japanese-wikipedia")

Reference

安岡孝一: 青空文庫DeBERTaモデルによる国語研長単位係り受け解析, 東洋学へのコンピュータ利用, 第35回研究セミナー (2022年7月), pp.29-43.