xnli、xnli

参考文献:

あーる

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/ar')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 既知のライセンスはありません
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

バックグラウンド

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/bg')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 既知のライセンスはありません
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

デ

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/de')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 既知のライセンスはありません
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

エル

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/el')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 既知のライセンスはありません
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

jp

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/en')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 既知のライセンスはありません
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

エス

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/es')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 不明なライセンス
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

フランス

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/fr')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 不明なライセンス
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

こんにちは

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/hi')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 既知のライセンスはありません
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

る

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/ru')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 既知のライセンスはありません
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

スイス

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/sw')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 既知のライセンスはありません
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

番目

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/th')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 既知のライセンスはありません
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

tr

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/tr')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 不明なライセンス
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

あなた

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/ur')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 不明なライセンス
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

ヴィ

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/vi')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 不明なライセンス
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

zh

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/zh')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 不明なライセンス
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴： 
{
    "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"
    }
}

すべての言語

次のコマンドを使用して、このデータセットを TFDS にロードします。

ds = tfds.load('huggingface:xnli/all_languages')
  • 説明：
XNLI is a subset of a few thousand examples from MNLI which has been translated
into a 14 different languages (some low-ish resource). As with MNLI, the goal is
to predict textual entailment (does sentence A imply/contradict/neither sentence
B) and is a classification task (given two sentences, predict one of three
labels).
  • ライセンス: 不明なライセンス
  • バージョン: 1.1.0
  • 分割:
スプリット例
'test' 5010
'train' 392702
'validation' 2490
  • 特徴：
{
    "premise": {
        "languages": [
            "ar",
            "bg",
            "de",
            "el",
            "en",
            "es",
            "fr",
            "hi",
            "ru",
            "sw",
            "th",
            "tr",
            "ur",
            "vi",
            "zh"
        ],
        "id": null,
        "_type": "Translation"
    },
    "hypothesis": {
        "languages": [
            "ar",
            "bg",
            "de",
            "el",
            "en",
            "es",
            "fr",
            "hi",
            "ru",
            "sw",
            "th",
            "tr",
            "ur",
            "vi",
            "zh"
        ],
        "num_languages": 15,
        "id": null,
        "_type": "TranslationVariableLanguages"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}