- Deskripsi :
Race adalah kumpulan data pemahaman bacaan skala besar dengan lebih dari 28.000 bagian dan hampir 100.000 pertanyaan. Dataset dikumpulkan dari ujian bahasa Inggris di Cina, yang dirancang untuk siswa sekolah menengah dan sekolah menengah atas. Dataset dapat disajikan sebagai set pelatihan dan pengujian untuk pemahaman mesin.
Deskripsi konfigurasi : Konfigurasi pembuat untuk kumpulan data RACE.
Situs web : https://www.cs.cmu.edu/~glai1/data/race/
Kode sumber :
tfds.text.race.Race
Versi :
-
1.0.0
: Rilis awal. -
2.0.0
(default): Tambahkan contoh id.
-
Ukuran unduhan :
24.26 MiB
Cache otomatis ( dokumentasi ): Ya
Fitur :
FeaturesDict({
'answers': Sequence(Text(shape=(), dtype=tf.string)),
'article': Text(shape=(), dtype=tf.string),
'example_id': Text(shape=(), dtype=tf.string),
'options': Sequence(Sequence(Text(shape=(), dtype=tf.string))),
'questions': Sequence(Text(shape=(), dtype=tf.string)),
})
Kunci yang diawasi (Lihat
as_supervised
doc ):None
Kutipan :
@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}
}
- Gambar ( tfds.show_examples ): Tidak didukung.
ras / tinggi (konfigurasi default)
Ukuran
52.39 MiB
data :52.39 MiB
Split :
Membagi | Contoh |
---|---|
'dev' | 1.021 |
'test' | 1.045 |
'train' | 18.728 |
- Contoh ( tfds.as_dataframe ):
ras / menengah
Ukuran
12.51 MiB
data :12.51 MiB
Split :
Membagi | Contoh |
---|---|
'dev' | 368 |
'test' | 362 |
'train' | 6.409 |
- Contoh ( tfds.as_dataframe ):