metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: start_positions
dtype: int64
- name: end_positions
dtype: int64
splits:
- name: train
num_bytes: 172176192
num_examples: 88568
- name: validation
num_bytes: 20975760
num_examples: 10790
download_size: 27919691
dataset_size: 193151952
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*