Update tokenization_qwen.py

tokenization_qwen.py

@@ -394,7 +394,7 @@ class QWenTokenizer(PreTrainedTokenizer):
         for ele in list_format:
             if 'image' in ele:
                 num_images += 1
-                text += f'Picture {num_images}:'
+                text += f'Picture {num_images}: '
                 text += self.image_start_tag + ele['image'] + self.image_end_tag
                 text += '\n'
             elif 'text' in ele: