기가워드

참조:

다음 명령을 사용하여 TFDS에서 이 데이터세트를 로드합니다.

ds = tfds.load('huggingface:gigaword')
  • 설명 :
Headline-generation on a corpus of article pairs from Gigaword consisting of
around 4 million articles. Use the 'org_data' provided by
https://github.com/microsoft/unilm/ which is identical to
https://github.com/harvardnlp/sent-summary but with better format.

There are two features:
  - document: article.
  - summary: headline.
  • 라이선스 : 알려진 라이선스 없음
  • 버전 : 1.2.0
  • 분할 :
나뉘다
'test' 1951년
'train' 3803957
'validation' 189651
  • 특징 :
{
    "document": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "summary": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}