Update app.py

app.py

@@ -73,82 +73,32 @@ def get_file(source_documents):
 
 
 def chat_query_doc(question, chat_history_doc):
-    
-    query_old = f"""Provide an elaborate, precise and pointwise reply to the question: {question}.
-                    Also, Please consider the provided chat history: {chat_history_doc}.
-                    Ensure that your current response is detailed, accurate, and addresses each aspect of the question thoroughly.
-                    If the context of the question doesn't align with your last reply, please provide your response in a fresh manner.
-                    If don't get the answer, feel free to reply from your own knowledge."""
-    
-    
-    # query = f"""You'll be asked with a User Query. If the Query is related to Electrical Domain, Provide a precise and point-wise reply to the query: {question} \
-    #            based on provided context only. Ensure that your reply addresses each aspect of the query thoroughly. """
 
-    query = f""" Provide a precise and point-wise reply to the query: {question} \
-                based on provided context only. Ensure that your reply addresses each aspect of the query thoroughly, and highlight the important points using text formatting 
-                in your reply."""
+    query = f"""Please provide a precise, point-wise reply to the query: {question}.\
+                Highlight the important points using properly formatted text, such as bullet points, bold text, or italics where appropriate."""
 
-    retriever = vectordb.as_retriever()
-    #repo_id = "mistralai/Mistral-7B-Instruct-v0.2"
-    repo_id="HuggingFaceH4/zephyr-7b-beta"
-    
-    # ChatVectorDBChain_24.10.2024
     #llm = OllamaLLM(model="llama3")
     #llm = Ollama(model="llama3")
-       
-    llm = HuggingFaceEndpoint(repo_id="HuggingFaceH4/zephyr-7b-beta", temperature = 0.5, huggingfacehub_api_token=hf_token,)
-
-    qa_template = """You are an AI assistant for answering questions. You are given the following extracted parts of a long document and a question. 
-                    Provide a conversational answer. If you don't know the answer, just say "Hmm, I'm not sure.".
-                    Don't try to make up an answer. 
-                    Question: {question}
-                    =========
-                    {context}
-                    =========
-                    Answer in Markdown:"""
-
-    
-    qa_prompt = PromptTemplate(template=qa_template, input_variables=["question", "context"])
-
-    llm_chain = qa_prompt | llm
-        
-    qa_chain = ConversationalRetrievalChain.from_llm(
-        llm,
-        retriever = retriever, 
-        return_source_documents = True
-    )
-    
-    result = qa_chain({"question": question, "chat_history" : chat_history_doc})
-    chat_history_doc.append((question, result["answer"]))
-    
+    #repo_id = "mistralai/Mistral-7B-Instruct-v0.2"
     #llm = ChatOpenAI(model = llm_name, temperature = 0.1, api_key = OPENAI_API_KEY)
-    #llm = GoogleGenerativeAI(model = "gemini-pro", google_api_key = GEMINI_API_KEY)                                   ###
+    #llm = GoogleGenerativeAI(model = "gemini-pro", google_api_key = GEMINI_API_KEY)
     #llm = ChatGoogleGenerativeAI(model = "gemini-1.0-pro", google_api_key = GEMINI_API_KEY, temperature = 0)
 
-    # # Conversation Retrival Chain with Memory
-    # #memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
-    # retriever = vectordb.as_retriever()
-    # qa = ConversationalRetrievalChain.from_llm(llm, retriever = retriever, return_source_documents = True)
-
-
-     
-
-    # Replace input() with question variable for Gradio
-    # result = qa({"question": query, "chat_history" : chat_history_doc})
+    llm = HuggingFaceEndpoint(repo_id="HuggingFaceH4/zephyr-7b-beta", temperature=0.5, huggingfacehub_api_token=HF_token)
 
-    # Update the history with the latest question and response
-    # history.append({"user": question, "bot": result["answer"]})
-    # chat_history_doc.append((query, result["answer"]))
+    retriever = vectordb.as_retriever()
+    memory_doc = ConversationBufferMemory(memory_key="chat_history", return_messages=True, output_key="answer")
+    qa = ConversationalRetrievalChain.from_llm(llm, retriever=retriever, return_source_documents=True, memory=memory_doc) 
+    result = qa({"question": query, "chat_history": chat_history_doc})
+    chat_history_doc.append((question, result["answer"]))
 
     source_docs = result["source_documents"]
     file_names = get_file(source_docs)
-    # file_name = os.path.basename(source_docs[0].metadata['source'])
     file_name = ', '.join([f"{x}" for x in file_names[:3]])
 
-    # print("History : ", history)
-    # print("\n Chat_his : ", chat_history)
-
     return result["answer"] + "\n\nSources : " + file_name
+  
+
 
 
 def chat_query_IS(question, chat_history_IS):