metadata
dataset_info:
features:
- name: pairs
dtype: string
- name: scores
dtype: float64
splits:
- name: train
num_bytes: 2717831656
num_examples: 70107704
download_size: 824884491
dataset_size: 2717831656
configs:
- config_name: default
data_files:
- split: train
path: data/train-*