Attend the Women in ML Symposium on December 7 Register now

kor_nli

References:

multi_nli

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:kor_nli/multi_nli')
  • Description:
Korean Natural  Language Inference datasets
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'train' 392702
  • Features:
{
    "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

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:kor_nli/snli')
  • Description:
Korean Natural  Language Inference datasets
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'train' 550152
  • Features:
{
    "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

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:kor_nli/xnli')
  • Description:
Korean Natural  Language Inference datasets
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 5010
'validation' 2490
  • Features:
{
    "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"
    }
}