metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 27564311544
num_examples: 21400863
- name: validation
num_bytes: 3008263104
num_examples: 2335608
download_size: 7092165713
dataset_size: 30572574648