LCCC

参考文献:

大きい

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:lccc/large')
  • 説明
LCCC: Large-scale Cleaned Chinese Conversation corpus (LCCC) is a large corpus of Chinese conversations.
A rigorous data cleaning pipeline is designed to ensure the quality of the corpus.
This pipeline involves a set of rules and several classifier-based filters.
Noises such as offensive or sensitive words, special symbols, emojis,
grammatically incorrect sentences, and incoherent conversations are filtered.
  • ライセンス: MIT
  • バージョン: 1.0.0
  • 分割:
スプリット
'train' 12007759
  • 特徴
{
    "dialog": [
        {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        }
    ]
}

ベース

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:lccc/base')
  • 説明
LCCC: Large-scale Cleaned Chinese Conversation corpus (LCCC) is a large corpus of Chinese conversations.
A rigorous data cleaning pipeline is designed to ensure the quality of the corpus.
This pipeline involves a set of rules and several classifier-based filters.
Noises such as offensive or sensitive words, special symbols, emojis,
grammatically incorrect sentences, and incoherent conversations are filtered.
  • ライセンス: MIT
  • バージョン: 1.0.0
  • 分割:
スプリット
'test' 10000
'train' 6820506
'validation' 20000
  • 特徴
{
    "dialog": [
        {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        }
    ]
}