dataset:
  name: Mixed Text Dataset
  total_samples: 20000

sources:
  - name: Wikipedia
    samples: 10000
    details: "20220301 dump"

  - name: Gutenberg
    samples: 3000
    details: "GutenDex API"

  - name: CNN/DailyMail
    samples: 7000
    details: "Version 3.0.0"

preprocessing:
  method: regex_cleaning
  description: "Replaced patterns like (.*?/) with whitespace"