cosmo_qa

  • Descrizione :

Cosmos QA è un set di dati su larga scala di 35,6K problemi che richiedono una comprensione della lettura basata sul buon senso, formulati come domande a scelta multipla. Si concentra sulla lettura tra le righe di una raccolta diversificata di narrazioni quotidiane delle persone, ponendo domande riguardanti le probabili cause o effetti di eventi che richiedono un ragionamento al di là delle esatte estensioni del testo nel contesto.

Diviso Esempi
'test' 6.963
'train' 25.262
'validation' 2.985
  • Struttura delle caratteristiche :
FeaturesDict({
    'answer0': Text(shape=(), dtype=string),
    'answer1': Text(shape=(), dtype=string),
    'answer2': Text(shape=(), dtype=string),
    'answer3': Text(shape=(), dtype=string),
    'context': Text(shape=(), dtype=string),
    'id': Text(shape=(), dtype=string),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=4),
    'question': Text(shape=(), dtype=string),
})
  • Documentazione delle funzionalità :
Caratteristica Classe Forma Tipo D Descrizione
CaratteristicheDict
risposta0 Testo corda
risposta 1 Testo corda
risposta2 Testo corda
risposta3 Testo corda
contesto Testo corda
id Testo corda
etichetta ClassLabel int64
domanda Testo corda
  • Citazione :
@inproceedings{huang-etal-2019-cosmos,
    title = "Cosmos {QA}: Machine Reading Comprehension with Contextual Commonsense Reasoning",
    author = "Huang, Lifu  and
      Le Bras, Ronan  and
      Bhagavatula, Chandra  and
      Choi, Yejin",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)",
    year = "2019",
    url = "https://www.aclweb.org/anthology/D19-1243"
}