Update generate_reference_files.py

generate_reference_files.py

@@ -0,0 +1,16 @@
+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('lmqg/qag_tweetqa', split=split, download_mode='force_redownload')
+    for data in ['questions_answers', '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)}"