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: 123617016
num_examples: 63589
download_size: 16376332
dataset_size: 123617016