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: 170292456
num_examples: 87599
- name: validation
num_bytes: 20548080
num_examples: 10570
download_size: 26850313
dataset_size: 190840536
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*