README.md

metadata

dataset_info:
  - config_name: default
    features:
      - name: text
        dtype: string
      - name: summary
        dtype: string
      - name: id
        dtype: string
    splits:
      - name: train
        num_bytes: 1260614391
        num_examples: 287113
      - name: validation
        num_bytes: 57672089
        num_examples: 13368
      - name: test
        num_bytes: 49877597
        num_examples: 11490
    download_size: 836881022
    dataset_size: 1368164077
  - config_name: mini
    features:
      - name: text
        dtype: string
      - name: summary
        dtype: string
      - name: id
        dtype: string
      - name: cluster_id
        dtype: int64
    splits:
      - name: train
        num_bytes: 169001211
        num_examples: 38500
    download_size: 103090954
    dataset_size: 169001211
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*
  - config_name: mini
    data_files:
      - split: train
        path: mini/train-*
license: odc-by
task_categories:
  - summarization
language:
  - en

cnn_dailymail: cleaned

Original cnn_dailymail config 3.0.0 with the following changes:

• renamed columns to text and summary
• cleaning applied to summary column to correct punctuation, etc.