Skip to content

vocab 생성관련 mecab pretokenize 방식 #43

Description

@soonilbae

vocab 생성관련 closed된 issue들 확인했습니다. 답변하신대로 KLUE Paper 4.1부분을
확인하였는데, Mecab으로 pretokenize한다는 것이 전체 코퍼스를 단순히 형태소 단위로만
분리하여 wordpiece로 학습한다는 의미인가요?

논문의 예제를 보니 단순히 형태소 단위로 분리하는 것 이상의 전처리가 필요한
것이 아닌가하는 생각이 들어서요. 초보자이다 보니 너무 간단한 질문을 하게
되는 것 같습니다. 감사합니다.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    questionFurther information is requested

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions