metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: token_type_ids
sequence: int8
- name: attention_mask
sequence: int8
- name: special_tokens_mask
sequence: int8
splits:
- name: train
num_bytes: 38003435808
num_examples: 2647954
download_size: 10083076260
dataset_size: 38003435808
configs:
- config_name: default
data_files:
- split: train
path: data/train-*