opus_tedtalks

参考文献:

毎時

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:opus_tedtalks/en-hr')
  • 説明
This is a Croatian-English parallel corpus of transcribed and translated TED talks, originally extracted from https://wit3.fbk.eu. The corpus is compiled by Željko Agić and is taken from http://lt.ffzg.hr/zagic provided under the CC-BY-NC-SA license.
2 languages, total number of files: 2
total number of tokens: 2.81M
total number of sentence fragments: 0.17M
  • ライセンス: 既知のライセンスはありません
  • バージョン: 1.0.0
  • 分割:
スプリット
'train' 86348
  • 特徴
{
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "languages": [
            "en",
            "hr"
        ],
        "id": null,
        "_type": "Translation"
    }
}