kor_nli

Referencias:

multi_nli

Utilice el siguiente comando para cargar este conjunto de datos en TFDS:

ds = tfds.load('huggingface:kor_nli/multi_nli')
  • Descripción :
Korean Natural  Language Inference datasets
  • Licencia : Ninguna licencia conocida
  • Versión : 1.0.0
  • Divisiones :
Dividir Ejemplos
'train' 392702
  • Características :
{
    "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

Utilice el siguiente comando para cargar este conjunto de datos en TFDS:

ds = tfds.load('huggingface:kor_nli/snli')
  • Descripción :
Korean Natural  Language Inference datasets
  • Licencia : Ninguna licencia conocida
  • Versión : 1.0.0
  • Divisiones :
Dividir Ejemplos
'train' 550152
  • Características :
{
    "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

Utilice el siguiente comando para cargar este conjunto de datos en TFDS:

ds = tfds.load('huggingface:kor_nli/xnli')
  • Descripción :
Korean Natural  Language Inference datasets
  • Licencia : Ninguna licencia conocida
  • Versión : 1.0.0
  • Divisiones :
Dividir Ejemplos
'test' 5010
'validation' 2490
  • Características :
{
    "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"
    }
}