Update QnA.py

QnA.py

@@ -1,6 +1,7 @@
 from langchain.chains.combine_documents import create_stuff_documents_chain
 from langchain_core.prompts import ChatPromptTemplate
 from langchain.chains import create_retrieval_chain
+from langchain.chains.summarize.chain import load_summarize_chain
 
 #from Api_Key import google_plam
 from langchain_groq import ChatGroq
@@ -11,14 +12,31 @@ load_dotenv()
 
 def prompt_template_to_analyze_resume():
     template = """
-    You are provided with the Resume of the Candidate in the context below . As an Talent Aquistion bot , your task is to provide insights about the candidate . 
-    If and only if asked about reliability , check How frequently the candidate has switched from one company to another. 
+    You are provided with the Resume of the Candidate in the context below . As an Talent Aquistion bot , your task is to provide insights about the 
+    candidate in point wise. Mention his strength and wekaness in general.Do not make up answers.
+    
+    \n\n:{context}
+    """
+    prompt = ChatPromptTemplate.from_messages(
+        [
+            ('system',template),
+            ('human','input'),
+        ]
+        )
+    return prompt
+
+def prompt_template_for_relaibility():
+    template ="""
+    You are provided with the Resume of the Candidate in the context below
+    If asked about reliability , check How frequently the candidate has switched from one company to another. 
     Grade him on the given basis: 
         If less than 2 Year - very less Reliable  
         if more than 2 years but less than 5 years - Reliable 
         if more than 5 Years - Highly Reliable
+    and generate verdict . 
     
     \n\n:{context}
+    
     """
     prompt = ChatPromptTemplate.from_messages(
         [
@@ -27,13 +45,26 @@ def prompt_template_to_analyze_resume():
         ]
         )
     return prompt
+    
 
-def Q_A(vectorstore,question,API_KEY):
+def summarize(documents,llm):
+    summarize_chain = load_summarize_chain(llm=llm, chain_type='refine', verbose = True)
+    results = summarize_chain.invoke({'input_documents':documents})
+    return results['output_text']
+
+
+def Q_A(vectorstore,documents,question,API_KEY):
     os.environ["GROQ_API_KEY"] = API_KEY
     llm_groq = ChatGroq(model="llama3-8b-8192")
+    
     # Create a retriever
     retriever = vectorstore.as_retriever(search_type = 'similarity',search_kwargs = {'k':2},)
-    question_answer_chain = create_stuff_documents_chain(llm_groq, prompt_template_to_analyze_resume())
+    if  'relaible' in question.lower() or 'relaibility' in question.lower():
+        question_answer_chain = create_stuff_documents_chain(llm_groq, prompt_template_for_relaibility())
+        
+    else:
+        question_answer_chain = create_stuff_documents_chain(llm_groq, prompt_template_to_analyze_resume())
+    
     chain = create_retrieval_chain(retriever, question_answer_chain)
     result = chain.invoke({'input':question})
     return result['answer']