README.md

metadata

dataset_info:
  features:
    - name: context
      dtype: string
    - name: question
      dtype: string
    - name: answers
      sequence: string
    - name: key
      dtype: string
    - name: labels
      list:
        - name: end
          sequence: int64
        - name: start
          sequence: int64
  splits:
    - name: train
      num_bytes: 100706304
      num_examples: 104071
    - name: validation
      num_bytes: 12941478
      num_examples: 12836
  download_size: 61870589
  dataset_size: 113647782

Dataset Card for "naturalquestionsshortqa"

Split taken from the MRQA 2019 Shared Task, formatted and filtered for Question Answering. For the original dataset, have a look here.