Referencias:
Utilice el siguiente comando para cargar este conjunto de datos en TFDS:
ds = tfds.load('huggingface:gigaword')
- Descripción :
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.
- Licencia : Sin licencia conocida
- Versión : 1.2.0
- Divisiones :
| Separar | Ejemplos |
|---|---|
'test' | 1951 |
'train' | 3803957 |
'validation' | 189651 |
- Características :
{
"document": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"summary": {
"dtype": "string",
"id": null,
"_type": "Value"
}
}