Datasets:
lmqg
/

Languages:
Italian
ArXiv:
License:
qg_itquad / generate_reference_files.py
asahi417's picture
update
2f6faf4
raw
history blame
827 Bytes
import os
from glob import glob
from datasets import load_dataset
os.makedirs('./reference_files', exist_ok=True)
for split in ['validation', 'test']:
dataset = load_dataset('asahi417/qg_itquad', split=split, download_mode='force_redownload')
for data in ['question', 'answer', 'sentence', 'paragraph']:
with open('./reference_files/{}-{}.txt'.format(data, split), 'w') as f:
if data == 'paragraph':
tmp_data = dataset['paragraph_id']
else:
tmp_data = dataset[data]
f.write('\n'.join([i.replace('\n', '.') for i in tmp_data]))
length = [len(open(i).read().split('\n')) for i in glob(f'reference_files/*{split}.txt')]
assert len(list(set(length))) == 1, length
assert length[0] == len(dataset), f"{length[0]} != {len(dataset)}"