metadata

dataset_info:
  features:
    - name: category
      dtype: string
    - name: text
      dtype: string
    - name: label
      dtype: int64
  splits:
    - name: train
      num_bytes: 23246239
      num_examples: 9245
    - name: test
      num_bytes: 2587727
      num_examples: 1028
  download_size: 16334175
  dataset_size: 25833966
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*
license: mit
language:
  - de
pretty_name: 10kGNAD - Ten Thousand German News Articles Dataset
size_categories:
  - 10M<n<100M
task_categories:
  - text-classification
tags:
  - News

10kGNAD - Ten Thousand German News Articles Dataset

German news articels, categorized. Category is the actual category name, one of: 'Etat', 'Inland', 'International', 'Kultur', 'Panorama', 'Sport', 'Web', 'Wirtschaft', 'Wissenschaft'. The label is the index as an integer.

DatasetDict({
    train: Dataset({
        features: ['category', 'text', 'label'],
        num_rows: 9245
    })
    test: Dataset({
        features: ['category', 'text', 'label'],
        num_rows: 1028
    })
})

Huggingface datasets version of https://github.com/tblock/10kGNAD