small fix to llm extraction file (#162)

Author: vibhav-vineet

eureka_ml_insights/user_configs/llm_extraction.py

@@ -149,7 +149,7 @@ def configure_subpipeline(
                     # consolidate new {extracted_answer_col} to replace the original {extracted_answer_col} whenever the latter is empty
                     # the initial if statement checks whether there has been a join beforehand
                     "transform": RunPythonTransform(
-                        f"df['{extracted_answer_col}'] = df.apply(lambda row: row['{extracted_answer_col}'] if '{extracted_answer_col}_x' not in row else row['{extracted_answer_col}_y'] if row['{extracted_answer_col}_x'] == '' else row['{extracted_answer_col}_x'], axis=1)"
+                        f"df['{extracted_answer_col}'] = df.apply(lambda row: row['{extracted_answer_col}'] if '{extracted_answer_col}_x' not in row else row['{extracted_answer_col}_y'] if row['{extracted_answer_col}_x'] == '{not_extracted_answer_value}' else row['{extracted_answer_col}_x'], axis=1)"
                     ),
                 },
             ),