人種

  • 説明:

Race は、28,000 を超えるパッセージと 100,000 近くの問題を含む大規模な読解データセットです。このデータセットは、中学生と高校生を対象とした中国の英語試験から収集されています。データセットは、機械理解のトレーニング セットとテスト セットとして使用できます。

FeaturesDict({
   
'answers': Sequence(Text(shape=(), dtype=string)),
   
'article': Text(shape=(), dtype=string),
   
'example_id': Text(shape=(), dtype=string),
   
'options': Sequence(Sequence(Text(shape=(), dtype=string))),
   
'questions': Sequence(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
答えシーケンス(テキスト) (なし、)ストリング
記事文章ストリング
example_id文章ストリング
オプションSequence(シーケンス(テキスト)) (なし、なし)ストリング
質問シーケンス(テキスト) (なし、)ストリング
@article{lai2017large,
    title
={RACE: Large-scale ReAding Comprehension Dataset From Examinations},
    author
={Lai, Guokun and Xie, Qizhe and Liu, Hanxiao and Yang, Yiming and Hovy, Eduard},
    journal
={arXiv preprint arXiv:1704.04683},
    year
={2017}
}

race/high (デフォルト設定)

  • データセットのサイズ: 52.39 MiB

  • スプリット:

スプリット
'dev' 1,021
'test' 1,045
'train' 18,728

レース/ミドル

  • データセットサイズ: 12.51 MiB

  • スプリット:

スプリット
'dev' 368
'test' 362
'train' 6,409