README.md

metadata

dataset_info:
  features:
    - name: qas
      list:
        - name: answers
          list:
            - name: answer_start
              dtype: int64
            - name: element_id
              dtype: int64
            - name: text
              dtype: string
        - name: id
          dtype: string
        - name: question
          dtype: string
    - name: page_id
      dtype: string
    - name: Image
      dtype: image
    - name: domain
      dtype: string
  splits:
    - name: train
      num_bytes: 22537566
      num_examples: 497
    - name: validation
      num_bytes: 4140875
      num_examples: 97
  download_size: 23191386
  dataset_size: 26678441
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
license: cc-by-4.0
task_categories:
  - question-answering
language:
  - en

For more details, please refer to paper WebSRC: A Dataset for Web-Based Structural Reading Comprehension.