metadata
dataset_info:
features:
- name: input_ids
sequence: int64
- name: attention_mask
sequence: int64
- name: answer
dtype: string
- name: start_positions
dtype: int64
- name: end_positions
dtype: int64
- name: token_type_ids
sequence: int64
splits:
- name: train
num_bytes: 17537973
num_examples: 1413
download_size: 1417570
dataset_size: 17537973
configs:
- config_name: default
data_files:
- split: train
path: data/train-*