pretrain dataset

scripts/prepare_pretrain_dataset.py

@@ -84,7 +84,7 @@ def batch_iterator(name=None):
 
     # text
     if name in (None, 'yirenc/general_knowledge_boolean'):
-        for split in ['train', 'test']:
+        for split in ['train', 'validation']:
             dataset = load_dataset(name, split=split)
 
             for row in dataset: