kor_nli

Bibliografia:

multi_nli

Użyj następującego polecenia, aby załadować ten zestaw danych do TFDS:

ds = tfds.load('huggingface:kor_nli/multi_nli')
  • Opis :
Korean Natural  Language Inference datasets
  • Licencja : Brak znanej licencji
  • Wersja : 1.0.0
  • Podziały :
Podział Przykłady
'train' 392702
  • Cechy :
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

snli

Użyj następującego polecenia, aby załadować ten zestaw danych do TFDS:

ds = tfds.load('huggingface:kor_nli/snli')
  • Opis :
Korean Natural  Language Inference datasets
  • Licencja : Brak znanej licencji
  • Wersja : 1.0.0
  • Podziały :
Podział Przykłady
'train' 550152
  • Cechy :
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

xnli

Użyj następującego polecenia, aby załadować ten zestaw danych do TFDS:

ds = tfds.load('huggingface:kor_nli/xnli')
  • Opis :
Korean Natural  Language Inference datasets
  • Licencja : Brak znanej licencji
  • Wersja : 1.0.0
  • Podziały :
Podział Przykłady
'test' 5010
'validation' 2490
  • Cechy :
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}