README.md

metadata

dataset_info:
  features:
    - name: question
      dtype: string
    - name: score_j
      dtype: float64
    - name: score_k
      dtype: float64
    - name: response_j
      dtype: string
    - name: response_k
      dtype: string
    - name: magnitude
      dtype: float64
  splits:
    - name: train
      num_bytes: 15244108
      num_examples: 84720
  download_size: 9046088
  dataset_size: 15244108
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*