- Deskripsi :
QASC adalah kumpulan data jawaban pertanyaan dengan fokus pada komposisi kalimat. Ini terdiri dari 9.980 pertanyaan pilihan ganda 8 arah tentang sains sekolah dasar (8.134 kereta, 926 dev, 920 tes), dan dilengkapi dengan korpus dari 17 juta kalimat.
Situs web : https://allenai.org/data/qasc
Kode sumber :
tfds.question_answering.qasc.Qasc
Versi :
-
0.1.0
(default): Tidak ada catatan rilis.
-
Ukuran unduhan :
1.54 MiB
Ukuran set data :
6.61 MiB
Cache otomatis ( dokumentasi ): Ya
Split :
Membagi | Contoh |
---|---|
'test' | 920 |
'train' | 8.134 |
'validation' | 926 |
- Fitur :
FeaturesDict({
'answerKey': Text(shape=(), dtype=tf.string),
'choices': Sequence({
'label': Text(shape=(), dtype=tf.string),
'text': Text(shape=(), dtype=tf.string),
}),
'combinedfact': Text(shape=(), dtype=tf.string),
'fact1': Text(shape=(), dtype=tf.string),
'fact2': Text(shape=(), dtype=tf.string),
'formatted_question': Text(shape=(), dtype=tf.string),
'id': Text(shape=(), dtype=tf.string),
'question': Text(shape=(), dtype=tf.string),
})
Kunci yang diawasi (Lihat
as_supervised
doc ):None
Kutipan :
@article{allenai:qasc,
author = {Tushar Khot and Peter Clark and Michal Guerquin and Peter Jansen and Ashish Sabharwal},
title = {QASC: A Dataset for Question Answering via Sentence Composition},
journal = {arXiv:1910.11473v2},
year = {2020},
}
Gambar ( tfds.show_examples ): Tidak didukung.
Contoh ( tfds.as_dataframe ):