README.md

metadata

dataset_info:
  features:
    - name: text
      dtype: string
    - name: hypothesis
      dtype: string
    - name: labels
      dtype:
        class_label:
          names:
            '0': entailment
            '1': not_entailment
    - name: task_name
      dtype: string
    - name: label_text
      dtype: string
  splits:
    - name: train
      num_bytes: 315013288
      num_examples: 1018733
  download_size: 206032209
  dataset_size: 315013288
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

Dataset Card for "dataset_train_nli"

Dataset for training a universal classifier. Additional information and training code available here: https://github.com/MoritzLaurer/zeroshot-classifier