c4

参考文献:

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:c4/en')
  • 説明
A colossal, cleaned version of Common Crawl's web crawl corpus.

Based on Common Crawl dataset: "https://commoncrawl.org".

This is the processed version of Google's C4 dataset by AllenAI.
  • ライセンス: 不明なライセンス
  • バージョン: 0.0.0
  • 分割:
スプリット
'train' 364868892
'validation' 364608
  • 特徴
{
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "timestamp": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "url": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

en.noblocklist

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:c4/en.noblocklist')
  • 説明
A colossal, cleaned version of Common Crawl's web crawl corpus.

Based on Common Crawl dataset: "https://commoncrawl.org".

This is the processed version of Google's C4 dataset by AllenAI.
  • ライセンス: 不明なライセンス
  • バージョン: 0.0.0
  • 分割:
スプリット
'train' 393391519
'validation' 393226
  • 特徴
{
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "timestamp": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "url": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

リアルニュースのような

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:c4/realnewslike')
  • 説明
A colossal, cleaned version of Common Crawl's web crawl corpus.

Based on Common Crawl dataset: "https://commoncrawl.org".

This is the processed version of Google's C4 dataset by AllenAI.
  • ライセンス: 不明なライセンス
  • バージョン: 0.0.0
  • 分割:
スプリット
'train' 13799838
'validation' 13863
  • 特徴
{
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "timestamp": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "url": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

en.noclean

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:c4/en.noclean')
  • 説明
A colossal, cleaned version of Common Crawl's web crawl corpus.

Based on Common Crawl dataset: "https://commoncrawl.org".

This is the processed version of Google's C4 dataset by AllenAI.
  • ライセンス: 不明なライセンス
  • バージョン: 0.0.0
  • 分割:
スプリット
'train' 1063805381
'validation' 1065029
  • 特徴
{
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "timestamp": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "url": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}