Update ranking/prompt.py

ranking/prompt.py

@@ -1,11 +1,12 @@
-RANKING_PROMPT_TEMPLATE = """[INST] You are an expert data scientist helping a researcher select the best dataset for their project.
+RANKING_PROMPT_TEMPLATE = """You are an expert data scientist helping a researcher select the best dataset for their project.
 Review the user query and the candidate datasets provided.
 Score each dataset from 0.0 to 10.0 based on how well it matches the user query and intent.
 
 CRITICAL INSTRUCTIONS:
 1. Do not penalize datasets for having short descriptions.
-2. If the dataset name and general domain strongly match the user query, assign a high score between 9.0 and 10.0.
-3. Only assign low scores if the dataset is completely irrelevant to the query.
+2. Penalise datasets for missing description.
+3. If the dataset name and general domain strongly match the user query, assign a high score between 9.0 and 10.0.
+4. Only assign low scores if the dataset is completely irrelevant to the query.
 
 Provide a brief, 1 sentence justification for your score.
 
@@ -23,5 +24,4 @@ Output your response in the following JSON format ONLY:
         "llm_score": 9.5,
         "reasoning": "Brief explanation of why this dataset is a good fit."
     }}
-]
-[/INST]"""
+]"""