metadata
dataset_info:
features:
- name: dataset_name
dtype: string
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 45672318962
num_examples: 2645905
download_size: 3900530058
dataset_size: 45672318962
configs:
- config_name: default
data_files:
- split: train
path: data/train-*