cos_e

  • Mô tả :

Common Sense Explanations (CoS-E) cho phép các mô hình ngôn ngữ đào tạo tự động tạo ra các giải thích có thể được sử dụng trong quá trình đào tạo và suy luận trong khung Commonsense Auto-Generated Explanation (CAGE) mới.

Tách ra ví dụ
'train' 9,741
'validation' 1.221
  • Cấu trúc tính năng :
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),
})
  • Tài liệu tính năng :
Tính năng Lớp Hình dạng Dtype Sự mô tả
Tính năngDict
trừu tượng_explanation Chữ sợi dây
câu trả lời Chữ sợi dây
lựa chọn Trình tự (Văn bản) (Không có,) sợi dây
extractive_explanation Chữ sợi dây
Tôi Chữ sợi dây
câu hỏi Chữ sợi dây
  • trích dẫn :
@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"
}