README.md

metadata

dataset_info:
  features:
    - name: data_source
      dtype: string
    - name: prompt
      list:
        - name: content
          dtype: string
        - name: role
          dtype: string
    - name: ability
      dtype: string
    - name: reward_model
      struct:
        - name: extraction_method
          dtype: string
        - name: ground_truth
          dtype: string
        - name: key
          dtype: string
        - name: style
          dtype: string
    - name: extra_info
      struct:
        - name: contest_id
          dtype: string
        - name: idx
          dtype: string
        - name: question_source
          dtype: string
        - name: source
          dtype: string
        - name: var_type
          dtype: string
    - name: lower_pass_rate
      dtype: float64
    - name: lower_scores
      list: float64
    - name: upper_pass_rate
      dtype: float64
    - name: upper_scores
      list: float64
    - name: goldilocks
      dtype: bool
  splits:
    - name: train
      num_bytes: 155278407
      num_examples: 98473
  download_size: 150111427
  dataset_size: 155278407
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*