adjust evaluation prompt

utils.py

@@ -73,9 +73,10 @@ def get_evaluation_with_gpt(task_description, text, api_key=None):
     prompt = (
         f"Given the task: {task_description}, the provided answer is: {text}\n"
         f"Evaluate the answer using a scale from 0 to 10. Scores should reflect stricter quality thresholds:\n"
-        f"0-4: Below average. The answer includes some relevant ideas but has no valuable contribution.\n"
-        f"5-7: Fair to average quality. The answer is relevant, with some specific insights.\n"
-        f"8-10: Good to excellent quality. The answer is insightful, valuable, and actionable.\n"
+        f"0-4 indicates below-average quality with significant shortcomings.\n"
+        f"5-6 represents acceptable quality.\n"
+        f"7-8 signifies good quality, showing substantial thought.\n"
+        f"9-10 represents exceptional quality, with insightful, and highly concrete suggestions.\n\n"
         f"When evaluating, use the entire scoring range and avoid defaulting to mid-range scores.\n\n"
         f"Evaluate based on the criteria:\n\n"
         f"Novelty: The uniqueness and innovation of the ideas."