wiki_split

Referências:

• Código
• Abraçando cara

Use o seguinte comando para carregar este conjunto de dados no TFDS:

ds = tfds.load('huggingface:wiki_split')

• Descrição :

One million English sentences, each split into two sentences that together preserve the original meaning, extracted from Wikipedia 
Google's WikiSplit dataset was constructed automatically from the publicly available Wikipedia revision history. Although 
the dataset contains some inherent noise, it can serve as valuable training data for models that split or merge sentences.

• Licença : Nenhuma licença conhecida
• Versão : 0.1.0
• Divisões :

• Características :

{
    "complex_sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_sentence_1": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_sentence_2": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}