metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: token_type_ids
sequence: int8
- name: attention_mask
sequence: int8
- name: labels
sequence: int64
splits:
- name: train
num_bytes: 1190392538880
num_examples: 20753008
download_size: 307400657843
dataset_size: 1190392538880