README.md

metadata

dataset_info:
  - config_name: adversary
    features:
      - name: question
        dtype: string
      - name: answers
        sequence: string
      - name: ctxs
        list:
          - name: hasanswer
            dtype: bool
          - name: score
            dtype: float64
          - name: text
            dtype: string
          - name: title
            dtype: string
      - name: gpt_answer_sentence
        dtype: string
      - name: gpt_adv_sentence
        sequence: string
      - name: is_valid_adv_sentence
        dtype: bool
      - name: gpt_adv_passage
        sequence: string
      - name: is_valid_adv_passage
        dtype: bool
    splits:
      - name: train
        num_bytes: 28520709
        num_examples: 3610
    download_size: 16013125
    dataset_size: 28520709
  - config_name: adversary-sent
    features:
      - name: question
        dtype: string
      - name: answers
        sequence: string
      - name: gpt_answer_sentence
        dtype: string
      - name: gpt_adv_sentence
        sequence: string
      - name: is_valid_adv_sentence
        dtype: bool
      - name: gpt_adv_passage
        sequence: string
      - name: is_valid_adv_passage
        dtype: bool
      - name: ctxs
        list:
          - name: hasanswer
            dtype: bool
          - name: score
            dtype: float32
          - name: text
            dtype: string
    splits:
      - name: train
        num_bytes: 7841766
        num_examples: 3610
    download_size: 4333156
    dataset_size: 7841766
  - config_name: adversary_v2
    features:
      - name: question
        dtype: string
      - name: answers
        sequence: string
      - name: ctxs
        list:
          - name: hasanswer
            dtype: bool
          - name: score
            dtype: float64
          - name: text
            dtype: string
          - name: title
            dtype: string
      - name: gpt_answer_sentence
        dtype: string
      - name: gpt_adv_sentence
        sequence: string
      - name: is_valid_adv_sentence
        dtype: bool
      - name: gpt_adv_passage
        sequence: string
      - name: is_valid_adv_passage
        dtype: bool
    splits:
      - name: train
        num_bytes: 28520588
        num_examples: 3610
    download_size: 16014456
    dataset_size: 28520588
  - config_name: adversary_v3
    features:
      - name: question
        dtype: string
      - name: answers
        sequence: string
      - name: ctxs
        list:
          - name: hasanswer
            dtype: bool
          - name: score
            dtype: float64
          - name: text
            dtype: string
          - name: title
            dtype: string
      - name: gpt_answer_sentence
        dtype: string
      - name: gpt_adv_sentence
        sequence: string
      - name: is_valid_adv_sentence
        dtype: bool
      - name: gpt_adv_passage
        sequence: string
      - name: is_valid_adv_passage
        dtype: bool
    splits:
      - name: train
        num_bytes: 28520588
        num_examples: 3610
    download_size: 16014456
    dataset_size: 28520588
  - config_name: default
    features:
      - name: question
        dtype: string
      - name: answers
        sequence: string
      - name: ctxs
        list:
          - name: hasanswer
            dtype: bool
          - name: score
            dtype: float64
          - name: text
            dtype: string
          - name: title
            dtype: string
    splits:
      - name: test
        num_bytes: 23673906
        num_examples: 3610
    download_size: 13529716
    dataset_size: 23673906
configs:
  - config_name: adversary
    data_files:
      - split: train
        path: adversary/train-*
  - config_name: adversary-sent
    data_files:
      - split: train
        path: adversary-sent/train-*
  - config_name: adversary_v2
    data_files:
      - split: train
        path: adversary_v2/train-*
  - config_name: adversary_v3
    data_files:
      - split: train
        path: adversary_v3/train-*
  - config_name: default
    data_files:
      - split: test
        path: data/test-*