metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: special_tokens_mask
sequence: int8
splits:
- name: train
num_bytes: 86351663844
num_examples: 27999891
download_size: 23243344520
dataset_size: 86351663844