Datasets:
Tasks:
Text Generation
Modalities:
Text
Sub-tasks:
language-modeling
Languages:
Korean
Size:
10K<n<100K
ArXiv:
License:
File size: 830 Bytes
23d3c55 439f7cd 23d3c55 d003f76 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 |
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_koquad', 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)}"
|