xnli

  • 説明:

XNLI は MNLI からの数千の例のサブセットであり、14 の異なる言語に翻訳されています (低めのリソース)。 MNLI と同様に、目標はテキストの含意を予測すること (文 A は含意/矛盾/文 B のいずれでもないか) であり、分類タスクです (2 つの文が与えられた場合、3 つのラベルのうちの 1 つを予測します)。

スプリット
'test' 5,010
'validation' 2,490
  • 機能構造:
FeaturesDict({
   
'hypothesis': TranslationVariableLanguages({
       
'language': Text(shape=(), dtype=string),
       
'translation': Text(shape=(), dtype=string),
   
}),
   
'label': ClassLabel(shape=(), dtype=int64, num_classes=3),
   
'premise': Translation({
       
'ar': Text(shape=(), dtype=string),
       
'bg': Text(shape=(), dtype=string),
       
'de': Text(shape=(), dtype=string),
       
'el': Text(shape=(), dtype=string),
       
'en': Text(shape=(), dtype=string),
       
'es': Text(shape=(), dtype=string),
       
'fr': Text(shape=(), dtype=string),
       
'hi': Text(shape=(), dtype=string),
       
'ru': Text(shape=(), dtype=string),
       
'sw': Text(shape=(), dtype=string),
       
'th': Text(shape=(), dtype=string),
       
'tr': Text(shape=(), dtype=string),
       
'ur': Text(shape=(), dtype=string),
       
'vi': Text(shape=(), dtype=string),
       
'zh': Text(shape=(), dtype=string),
   
}),
})
  • 機能のドキュメント:
特徴クラスDtype説明
特徴辞書
仮説TranslationVariableLanguages
仮説/言語文章ストリング
仮説・翻訳文章ストリング
ラベルクラスラベルint64
前提翻訳
前提/ar文章ストリング
前提/背景文章ストリング
前提/で文章ストリング
前提/エル文章ストリング
前提/en文章ストリング
前提/es文章ストリング
前提/フランス文章ストリング
前提/こんにちは文章ストリング
前提/ル文章ストリング
前提/SW文章ストリング
前提/目文章ストリング
前提/TR文章ストリング
前提/ウル文章ストリング
前提/vi文章ストリング
前提/zh文章ストリング
  • 引用
@InProceedings{conneau2018xnli,
  author
= "Conneau, Alexis
                 and Rinott, Ruty
                 and Lample, Guillaume
                 and Williams, Adina
                 and Bowman, Samuel R.
                 and Schwenk, Holger
                 and Stoyanov, Veselin"
,
  title
= "XNLI: Evaluating Cross-lingual Sentence Representations",
  booktitle
= "Proceedings of the 2018 Conference on Empirical Methods
               in Natural Language Processing"
,
  year
= "2018",
  publisher
= "Association for Computational Linguistics",
  location
= "Brussels, Belgium",
}
,

  • 説明:

XNLI は MNLI からの数千の例のサブセットであり、14 の異なる言語に翻訳されています (低めのリソース)。 MNLI と同様に、目標はテキストの含意を予測すること (文 A は含意/矛盾/文 B のいずれでもないか) であり、分類タスクです (2 つの文が与えられた場合、3 つのラベルのうちの 1 つを予測します)。

スプリット
'test' 5,010
'validation' 2,490
  • 機能構造:
FeaturesDict({
   
'hypothesis': TranslationVariableLanguages({
       
'language': Text(shape=(), dtype=string),
       
'translation': Text(shape=(), dtype=string),
   
}),
   
'label': ClassLabel(shape=(), dtype=int64, num_classes=3),
   
'premise': Translation({
       
'ar': Text(shape=(), dtype=string),
       
'bg': Text(shape=(), dtype=string),
       
'de': Text(shape=(), dtype=string),
       
'el': Text(shape=(), dtype=string),
       
'en': Text(shape=(), dtype=string),
       
'es': Text(shape=(), dtype=string),
       
'fr': Text(shape=(), dtype=string),
       
'hi': Text(shape=(), dtype=string),
       
'ru': Text(shape=(), dtype=string),
       
'sw': Text(shape=(), dtype=string),
       
'th': Text(shape=(), dtype=string),
       
'tr': Text(shape=(), dtype=string),
       
'ur': Text(shape=(), dtype=string),
       
'vi': Text(shape=(), dtype=string),
       
'zh': Text(shape=(), dtype=string),
   
}),
})
  • 機能のドキュメント:
特徴クラスDtype説明
特徴辞書
仮説TranslationVariableLanguages
仮説/言語文章ストリング
仮説・翻訳文章ストリング
ラベルクラスラベルint64
前提翻訳
前提/ar文章ストリング
前提/背景文章ストリング
前提/で文章ストリング
前提/エル文章ストリング
前提/en文章ストリング
前提/es文章ストリング
前提/フランス文章ストリング
前提/こんにちは文章ストリング
前提/ル文章ストリング
前提/SW文章ストリング
前提/目文章ストリング
前提/TR文章ストリング
前提/ウル文章ストリング
前提/vi文章ストリング
前提/zh文章ストリング
  • 引用
@InProceedings{conneau2018xnli,
  author
= "Conneau, Alexis
                 and Rinott, Ruty
                 and Lample, Guillaume
                 and Williams, Adina
                 and Bowman, Samuel R.
                 and Schwenk, Holger
                 and Stoyanov, Veselin"
,
  title
= "XNLI: Evaluating Cross-lingual Sentence Representations",
  booktitle
= "Proceedings of the 2018 Conference on Empirical Methods
               in Natural Language Processing"
,
  year
= "2018",
  publisher
= "Association for Computational Linguistics",
  location
= "Brussels, Belgium",
}