metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 5012855784
num_examples: 1027228
download_size: 1757582258
dataset_size: 5012855784
configs:
- config_name: default
data_files:
- split: train
path: data/train-*