metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: start_positions
sequence: int64
- name: end_positions
sequence: int64
- name: global_attention_mask
sequence: int64
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 36495790
num_examples: 586
- name: validation
num_bytes: 4341131
num_examples: 65
download_size: 4313316
dataset_size: 40836921
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*