The viewer is disabled because this dataset repo requires arbitrary Python code execution. Please consider removing the loading script and relying on automated data support (you can use convert_to_parquet from the datasets library). If this is not possible, please open a discussion for direct help.

Dataset Card for "cosmos_qa"

Dataset Summary

Cosmos QA is a large-scale dataset of 35.6K problems that require commonsense-based reading comprehension, formulated as multiple-choice questions. It focuses on reading between the lines over a diverse collection of people's everyday narratives, asking questions concerning on the likely causes or effects of events that require reasoning beyond the exact text spans in the context

Supported Tasks and Leaderboards

More Information Needed


More Information Needed

Dataset Structure

Data Instances


  • Size of downloaded dataset files: 24.40 MB
  • Size of the generated dataset: 24.51 MB
  • Total amount of disk used: 48.91 MB

An example of 'validation' looks as follows.

This example was too long and was cropped:

    "answer0": "If he gets married in the church he wo nt have to get a divorce .",
    "answer1": "He wants to get married to a different person .",
    "answer2": "He wants to know if he does nt like this girl can he divorce her ?",
    "answer3": "None of the above choices .",
    "context": "\"Do i need to go for a legal divorce ? I wanted to marry a woman but she is not in the same religion , so i am not concern of th...",
    "id": "3BFF0DJK8XA7YNK4QYIGCOG1A95STE##3180JW2OT5AF02OISBX66RFOCTG5J7##A2LTOS0AZ3B28A##Blog_56156##q1_a1##378G7J1SJNCDAAIN46FM2P7T6KZEW2",
    "label": 1,
    "question": "Why is this person asking about divorce ?"

Data Fields

The data fields are the same among all splits.


  • id: a string feature.
  • context: a string feature.
  • question: a string feature.
  • answer0: a string feature.
  • answer1: a string feature.
  • answer2: a string feature.
  • answer3: a string feature.
  • label: a int32 feature.

Data Splits

name train validation test
default 25262 2985 6963

Dataset Creation

Curation Rationale

More Information Needed

Source Data

Initial Data Collection and Normalization

More Information Needed

Who are the source language producers?

More Information Needed


Annotation process

More Information Needed

Who are the annotators?

More Information Needed

Personal and Sensitive Information

More Information Needed

Considerations for Using the Data

Social Impact of Dataset

More Information Needed

Discussion of Biases

More Information Needed

Other Known Limitations

More Information Needed

Additional Information

Dataset Curators

More Information Needed

Licensing Information

As reported via email by Yejin Choi, the dataset is licensed under CC BY 4.0 license.

Citation Information

    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)",
    month = nov,
    year = "2019",
    address = "Hong Kong, China",
    publisher = "Association for Computational Linguistics",
    url = "",
    doi = "10.18653/v1/D19-1243",
    pages = "2391--2401",


Thanks to @patrickvonplaten, @lewtun, @albertvillanova, @thomwolf for adding this dataset.

Downloads last month

Models trained or fine-tuned on allenai/cosmos_qa