cos_e

  • Deskripsi :

Common Sense Explanations (CoS-E) memungkinkan model bahasa pelatihan untuk secara otomatis menghasilkan penjelasan yang dapat digunakan selama pelatihan dan inferensi dalam kerangka kerja Commonsense Auto-Generated Explanation (CAGE) baru.

Membelah Contoh
'train' 9.741
'validation' 1.221
  • Struktur fitur :
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),
})
  • Dokumentasi fitur :
Fitur Kelas Membentuk Dtype Keterangan
fiturDict
abstractive_explanation Teks rangkaian
menjawab Teks rangkaian
pilihan Urutan (Teks) (Tidak ada,) rangkaian
extractive_explanation Teks rangkaian
Indo Teks rangkaian
pertanyaan Teks rangkaian
  • Kutipan :
@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"
}