cosmos_qa

  • বর্ণনা :

Cosmos QA হল 35.6K সমস্যার একটি বৃহৎ-স্কেল ডেটাসেট যার জন্য কমনসেন্স-ভিত্তিক রিডিং কম্প্রিহেনশন প্রয়োজন, বহু-পছন্দের প্রশ্ন হিসাবে তৈরি করা হয়েছে। এটি মানুষের দৈনন্দিন আখ্যানগুলির একটি বৈচিত্র্যময় সংগ্রহের উপর লাইনের মধ্যে পড়ার উপর দৃষ্টি নিবদ্ধ করে, ঘটনাগুলির সম্ভাব্য কারণ বা প্রভাব সম্পর্কে প্রশ্ন জিজ্ঞাসা করে যার জন্য প্রেক্ষাপটে সঠিক পাঠ্যের বাইরে যুক্তির প্রয়োজন হয়।

বিভক্ত উদাহরণ
'test' ৬,৯৬৩
'train' 25,262
'validation' 2,985
  • বৈশিষ্ট্য গঠন :
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),
})
  • বৈশিষ্ট্য ডকুমেন্টেশন :
বৈশিষ্ট্য ক্লাস আকৃতি ডিটাইপ বর্ণনা
ফিচারসডিক্ট
উত্তর0 পাঠ্য স্ট্রিং
উত্তর 1 পাঠ্য স্ট্রিং
উত্তর২ পাঠ্য স্ট্রিং
উত্তর3 পাঠ্য স্ট্রিং
প্রসঙ্গ পাঠ্য স্ট্রিং
আইডি পাঠ্য স্ট্রিং
লেবেল ক্লাসলেবেল int64
প্রশ্ন পাঠ্য স্ট্রিং
  • উদ্ধৃতি :
@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"
}