pretrain dataset

scripts/prepare_pretrain_dataset.py

@@ -74,6 +74,9 @@ def batch_iterator(name=None):
         dataset = load_dataset(name, split='train')
 
         for row in dataset:
+            if not row['Question'] or not row['Answer']:
+                continue
+             
             yield row['Question'] + ' ' + row['Answer']
 
         del dataset