metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: start_positions
dtype: int64
- name: end_positions
dtype: int64
- name: sequence_ids
sequence: int64
splits:
- name: train
num_bytes: 388306912
num_examples: 57544
- name: test
num_bytes: 97083476
num_examples: 14387
download_size: 46592442
dataset_size: 485390388
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*