README.md

metadata

configs:
  - config_name: default
    data_files:
      - split: mnli
        path: data/mnli-*
      - split: fevernli
        path: data/fevernli-*
      - split: anli
        path: data/anli-*
      - split: wanli
        path: data/wanli-*
      - split: lingnli
        path: data/lingnli-*
      - split: xnli
        path: data/xnli-*
dataset_info:
  features:
    - name: premise
      dtype: string
    - name: hypothesis
      dtype: string
    - name: label
      dtype:
        class_label:
          names:
            '0': entailment
            '1': neutral
            '2': contradiction
    - name: language
      dtype: string
  splits:
    - name: mnli
      num_bytes: 75405050
      num_examples: 392702
    - name: fevernli
      num_bytes: 76336755
      num_examples: 196805
    - name: anli
      num_bytes: 64930916
      num_examples: 162865
    - name: wanli
      num_bytes: 17409074
      num_examples: 102885
    - name: lingnli
      num_bytes: 5868113
      num_examples: 29985
    - name: xnli
      num_bytes: 9825139
      num_examples: 37350
  download_size: 0
  dataset_size: 249775047

Dataset Card for "mnli_anli_fevernli_wanli_lingnli_xnli_train"

Train data in a harmonized format for multiple NLI datasets.