metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: labels
dtype: int64
splits:
- name: train
num_bytes: 768394512
num_examples: 592897
- name: val
num_bytes: 135600480
num_examples: 104630
download_size: 328096338
dataset_size: 903994992
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: val
path: data/val-*