- Açıklama :
Dolphin Math Word Problem veri seti (2015), https://www.microsoft.com/en-us/research/uploads/prod/2016/02//dolphin-sigmadolphin.datasets.pdf
Ana sayfa : https://www.microsoft.com/en-us/research/project/sigmadolphin-2/
Kaynak kodu :
tfds.text.dolphin_number_word.DolphinNumberWord
Sürümler :
-
0.0.1
: İlk sürüm. -
0.0.2
(varsayılan): RaggedTensor düzeltmesi. Denklemler ve Kaynaklar, boşluklarla sınırlandırılmış bileşenlerle tek bir dize olarak temsil edilir -
0.0.3
: Kaynaksız örnekler içeren uç durumu işlemek için mantık yeniden sunuldu.
-
İndirme boyutu :
280.42 KiB
Veri kümesi boyutu :
1.49 MiB
Otomatik önbelleğe alınmış ( belgeler ): Evet
Bölmeler :
Bölmek | Örnekler |
---|---|
'test' | 3.507 |
'train' | 864 |
- Özellik yapısı :
FeaturesDict({
'ans': Text(shape=(), dtype=tf.string),
'equations': Text(shape=(), dtype=tf.string),
'id': Text(shape=(), dtype=tf.string),
'index': tf.int32,
'sources': Text(shape=(), dtype=tf.string),
'text': Text(shape=(), dtype=tf.string),
})
- Özellik belgeleri :
Özellik | Sınıf | Şekil | Dtype | Tanım |
---|---|---|---|---|
ÖzelliklerDict | ||||
ans | Metin | tf.string | ||
denklemler | Metin | tf.string | ||
İD | Metin | tf.string | ||
indeks | tensör | tf.int32 | ||
kaynaklar | Metin | tf.string | ||
Metin | Metin | tf.string |
Denetlenen anahtarlar (bkz
as_supervised
doc ):('text', 'ans')
Şekil ( tfds.show_examples ): Desteklenmez.
Örnekler ( tfds.as_dataframe ):
- alıntı :
@inproceedings{inproceedings,
author = {Shi, Shuming and Wang, Yuehui and Lin, Chin-Yew and Liu, Xiaojiang and Rui, Yong},
year = {2015},
month = {09},
pages = {},
title = {Automatically Solving Number Word Problems by Semantic Parsing and Reasoning},
doi = {10.18653/v1/D15-1135}
}