metadata
dataset_info:
features:
- name: input_ids
sequence:
sequence: int32
- name: attention_mask
sequence:
sequence: int8
- name: start_positions
sequence: int64
- name: end_positions
sequence: int64
splits:
- name: train
num_bytes: 916965
num_examples: 127
download_size: 377932
dataset_size: 916965
configs:
- config_name: default
data_files:
- split: train
path: data/train-*