metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: token_type_ids
sequence: int8
- name: attention_mask
sequence: int8
- name: start_positions
dtype: int64
- name: end_positions
dtype: int64
splits:
- name: train
num_bytes: 206437968
num_examples: 88524
- name: validation
num_bytes: 25148288
num_examples: 10784
download_size: 27211600
dataset_size: 231586256
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*