/벤치마크, 평가, 보정 /jev-spam-eval

jev-spam-eval

Noul 질문으로 제로샷 스팸 필터링을 수행해 TF-IDF 기준선과 비교합니다.

이 페이지는 영문 목록에서 생성했습니다.

readme에서

Zero-shot email classification with TypeSafe’s Jev TypeSafe’s Jev reached 98.64% accuracy on a 5,733-email ham/spam/phishing test using written category definitions and email context, without task-specific fine-tuning or labeled examples in its requests. A TF-IDF logistic regression classifier trained on roughly 4,600 labeled messages per fold reached …

상세

섹션
벤치마크, 평가, 보정
owner
bitnovus
스타
0
포크
0
최근 커밋
2026-09-18
언어
Jupyter Notebook
라이선스
MIT

어떤 프로젝트인가

형태
데이터셋 또는 벤치마크
호스트 에이전트
단독 실행
대상
연구자
성숙도
문서
실측치를 보고합니다

이럴 때 유용합니다

가장 가까운 용도