Datasets:
Tasks:
Text Generation
Sub-tasks:
language-modeling
Languages:
Japanese
Multilinguality:
monolingual
Size Categories:
10K<n<100K
Source Datasets:
SkelterLabsInc/JaQuAD
ArXiv:
License:
| 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_jaquad', split=split, download_mode='force_redownload') | |
| for data in ['question', 'answer', 'sentence', 'paragraph']: | |
| with open('./reference_files/{}-{}.txt'.format(data, split), 'w') as f: | |
| 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)}" | |