Fix dataset when swt-filter = False

Author: nielstron

src/dataset.py

@@ -171,7 +171,7 @@ def load_swebench_dataset(name="princeton-nlp/SWE-bench", split="test", is_swt:
     if filter_swt:
         return [i for i in instances if i["instance_id"] not in instance_filter]
     else:
-        instances
+        return instances
 
 
 ### MARK - Patch Correction
@@ -183,4 +183,4 @@ def load_swebench_dataset(name="princeton-nlp/SWE-bench", split="test", is_swt:
 PATCH_HUNK_PATTERN = re.compile(
     r"\@\@\s+\-(\d+),(\d+)\s+\+(\d+),(\d+)\s+\@\@(.+?)(?=diff\ |\-\-\-\ a\/|\@\@\ \-|\Z)",
     re.DOTALL,
-)
+)