metadata
dataset_info:
features:
- name: phrase_a
dtype: string
- name: phrase_b
dtype: string
splits:
- name: train
num_bytes: 4553558
num_examples: 21568
download_size: 1490641
dataset_size: 4553558
configs:
- config_name: default
data_files:
- split: train
path: data/train-*