wiki_split

Tài liệu tham khảo:

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:wiki_split')
  • Sự miêu tả :
One million English sentences, each split into two sentences that together preserve the original meaning, extracted from Wikipedia 
Google's WikiSplit dataset was constructed automatically from the publicly available Wikipedia revision history. Although 
the dataset contains some inherent noise, it can serve as valuable training data for models that split or merge sentences.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'test' 5000
'train' 989944
'validation' 5000
  • Đặc trưng :
{
    "complex_sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_sentence_1": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_sentence_2": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}