resrer-nq / README.md
seonglae's picture
Upload dataset
63091d4
metadata
dataset_info:
  features:
    - name: document_text
      dtype: string
    - name: long_answer_candidates
      list:
        - name: end_token
          dtype: int64
        - name: start_token
          dtype: int64
        - name: top_level
          dtype: bool
    - name: question_text
      dtype: string
    - name: annotations
      list:
        - name: annotation_id
          dtype: float64
        - name: long_answer
          struct:
            - name: candidate_index
              dtype: int64
            - name: end_token
              dtype: int64
            - name: start_token
              dtype: int64
        - name: short_answers
          list:
            - name: end_token
              dtype: int64
            - name: start_token
              dtype: int64
        - name: yes_no_answer
          dtype: string
    - name: document_url
      dtype: string
    - name: example_id
      dtype: int64
    - name: long_answer_text
      dtype: string
    - name: short_answer_text
      dtype: string
    - name: split_id
      dtype: string
    - name: answer_exist_chunk
      dtype: bool
    - name: summarization_text
      dtype: string
  splits:
    - name: train
      num_bytes: 497862929.4517183
      num_examples: 55113
  download_size: 121306017
  dataset_size: 497862929.4517183
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*