wyciągać się

  • opis :

Common Sense Explanations (CoS-E) umożliwia trenowanie modeli językowych w celu automatycznego generowania wyjaśnień, które można wykorzystać podczas uczenia i wnioskowania w nowatorskiej strukturze Commonsense Auto-Generated Explanation (CAGE).

Rozdzielać Przykłady
'train' 9741
'validation' 1221
  • Struktura funkcji :
FeaturesDict({
    'abstractive_explanation': Text(shape=(), dtype=string),
    'answer': Text(shape=(), dtype=string),
    'choices': Sequence(Text(shape=(), dtype=string)),
    'extractive_explanation': Text(shape=(), dtype=string),
    'id': Text(shape=(), dtype=string),
    'question': Text(shape=(), dtype=string),
})
  • Dokumentacja funkcji :
Funkcja Klasa Kształt Typ D Opis
FunkcjeDict
streszczenie_wyjaśnienie Tekst strunowy
odpowiadać Tekst strunowy
wybory Sekwencja (tekst) (Nic,) strunowy
wydobycie_wyjaśnienie Tekst strunowy
ID Tekst strunowy
pytanie Tekst strunowy
  • Cytat :
@inproceedings{rajani2019explain,
     title = "Explain Yourself! Leveraging Language models for Commonsense Reasoning",
    author = "Rajani, Nazneen Fatema  and
      McCann, Bryan  and
      Xiong, Caiming  and
      Socher, Richard",
      year="2019",
    booktitle = "Proceedings of the 2019 Conference of the Association for Computational Linguistics (ACL2019)",
    url ="https://arxiv.org/abs/1906.02361"
}