metadata
dataset_info:
features:
- name: id
dtype: string
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 33279393797
num_examples: 22522740
download_size: 14020873232
dataset_size: 33279393797
configs:
- config_name: default
data_files:
- split: train
path: data/train-*