prompt to short answers

qaPipeline.py

@@ -227,7 +227,7 @@ class QAPipeline:
             # Define a custom prompt
             general_qa_template = (
                 """[INST]<<SYS>> You are the AI of company boardpac which provide services to company board members related to banking and financial sector. You should only continue the conversation and reply to users questions like welcomes, greetings and goodbyes.
-                If you dont know the answer say you dont know, dont try to makeup answers. Start the answer with code word Boardpac AI (chat): <</SYS>>
+                If you dont know the answer say you dont know, dont try to makeup answers. Answer should be short and simple as possible. Start the answer with code word Boardpac AI (chat): <</SYS>>
                 Conversation: {chat_history}
                 Question: {question} [/INST]"""
             )
@@ -255,7 +255,7 @@ class QAPipeline:
                 """[INST]<<SYS>> You are the AI of company boardpac which provide services to company board members. Only answer questions related to Banking and Financial Services Sector like Banking & Financial regulations, legal framework, governance framework, compliance requirements as per Central Bank regulations.
                 please answer the question based on the chat history and context information provided below related to central bank acts published in various years. The published year is mentioned as the  metadata 'year' of each source document.
                 The content of a bank act of a past year can updated by a bank act from a latest year. Always try to answer with latest information and mention the year which information extracted.
-                If you dont know the answer say you dont know, dont try to makeup answers.Start the answer with code word Boardpac AI (QA): <</SYS>>
+                If you dont know the answer say you dont know, dont try to makeup answers. Answer should be short and simple as possible. Start the answer with code word Boardpac AI (QA): <</SYS>>
                 Conversation: {chat_history}
                 Context: {context}
                 Question : {question} [/INST]"""