metadata
dataset_info:
features:
- name: premise
dtype: string
- name: hypothesis
dtype: string
- name: label_name
dtype: string
- name: subset
dtype: string
splits:
- name: train
num_bytes: 351587589
num_examples: 1594065
- name: test
num_bytes: 29289215
num_examples: 112013
download_size: 149650337
dataset_size: 380876804
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
This dataset is a mixture of the following NLI (Natural Language Inference) and TE (Textual Entailment) datasets:
- SNLI
- MNLI
- Fever
- SciTail
- PAWS
- VitaminC
To merge the datasets together, labels has been converted to "entailment" and "non-entailment" using an NLI format with features "premise" and "hypothesis".