Can I read 生成モデルにおける日本語の敬語・非敬語の理解と処理 on EtoBox?

生成モデルにおける日本語の敬語・非敬語の理解と処理 by Tae Young is a document available to read on EtoBox.

What is 生成モデルにおける日本語の敬語・非敬語の理解と処理 about?

This paper investigates how generative models, particularly GPT-2, process Japanese polite and non-polite language by analyzing specific neurons associated with these forms. The study evaluates the performance of keigo neurons as binary classifiers and explores their behavior with input sentences, providing insights for enhancing language-specialized models. The findings contribute to understanding the model

Author

Tae Young

Language

EN