add system prompt

example/backend/server.py

@@ -125,7 +125,9 @@ async def generate_response(request: QueryRequest):
         context, paper_objects = format_papers_context(relevant_papers, similarities)
         
         # Prepare the prompt with context
-        full_prompt = f"{context}\n\nQuestion: {request.message}\nAnswer:"
+        # full_prompt = f"{context}\n\nQuestion: {request.message}\nAnswer:"
+        system_prompt = "You are a helpful assistant on the topic of materials science. If the question is not related to the scientific papers and research, please answer the question as a general assistant. For the scientic research questions, please only answer the question using the given retrieved papers and their contents. Do not generate the paper title and contents which do not exist in the retrieved context."
+        full_prompt = f"{system_prompt}\n\n{context}\n\nQuestion: {request.message}\n\nAnswer:"
         
         # Generate response using Llama model
         inputs = llama_tokenizer(full_prompt, return_tensors="pt").to(llama_model.device)
@@ -169,4 +171,4 @@ async def generate_response(request: QueryRequest):
         raise HTTPException(status_code=500, detail=f"An error occurred: {str(e)}")
 
 if __name__ == "__main__":
-    uvicorn.run("server:app", host="0.0.0.0", port=8000, reload=True)
+    uvicorn.run("server:app", host="0.0.0.0", port=8000, reload=True)