Update app.py

app.py

@@ -282,7 +282,8 @@ if create_vectorstores:
         documents = chunks,
         vectorstore_name="Vit_All_HF_Embeddings"
     )
-    print("vector_store_HF:",vector_store_HF._collection.count(),"chunks.")
+    print("vector_store_HF:",vector_store_HF._collection.count(),"chunks.", prompt_template)
+    
     print("")
 
 """
@@ -745,6 +746,76 @@ follow_up_question = "plaese give more details about it, including its use cases
 chain.invoke({"question":follow_up_question})['answer'])
 """
 
+
+
+# It is not clear to me if this is running. If you take it out, it still provides answers but also using different promptTemplate
+
+def create_ConversationalRetrievalChain(
+    llm,condense_question_llm,
+    retriever,
+    chain_type= 'stuff',
+    language="english",
+    model_name='gemini-pro'
+    #model_name='gpt-3.5-turbo'
+):
+    """Create a ConversationalRetrievalChain.
+    First, it passes the follow-up question along with the chat history to an LLM which rephrases 
+    the question and generates a standalone query. 
+    This query is then sent to the retriever, which fetches relevant documents (context) 
+    and passes them along with the standalone question and chat history to an LLM to answer.
+    """
+    
+    # 1. Define the standalone_question prompt. 
+    # Pass the follow-up question along with the chat history to the `condense_question_llm`
+    # which rephrases the question and generates a standalone question.
+
+    standalone_question_prompt = PromptTemplate(
+        input_variables=['chat_history', 'question'], 
+        template="""Given the following conversation and a follow up question, 
+rephrase the follow up question to be a standalone question, in its original language.\n\n
+Chat History:\n{chat_history}\n
+Follow Up Input: {question}\n
+Standalone question:""")
+
+    # 2. Define the answer_prompt
+    # Pass the standalone question + the chat history + the context (retrieved documents) to the `LLM` wihch will answer
+    
+    answer_prompt = ChatPromptTemplate.from_template(answer_template(language=language))
+
+    # 3. Add ConversationSummaryBufferMemory for gpt-3.5, and ConversationBufferMemory for the other models
+    
+    memory = create_memory(model_name)
+
+    # 4. Create the ConversationalRetrievalChain
+
+    chain = ConversationalRetrievalChain.from_llm(
+        condense_question_prompt=standalone_question_prompt,
+        combine_docs_chain_kwargs={'prompt': answer_prompt},
+        condense_question_llm=condense_question_llm,
+
+        memory=memory,
+        retriever = base_retriever_HF,  #changed this
+        #retriever = retriever,
+        #llm=llm,    #changed this
+        llm=instantiate_LLM(
+                LLM_provider="Google",api_key=google_api_key,temperature=0.5,
+                model_name="gemini-pro"),
+        chain_type= chain_type,
+        verbose= False,
+        return_source_documents=True    
+    )
+
+    print("Conversational retriever chain created successfully!")
+    
+    return chain,memory
+
+
+
+
+
+
+
+
 def submit_message(prompt, prompt_template, temperature, max_tokens, context_length, state):
     
     
@@ -754,6 +825,8 @@ def submit_message(prompt, prompt_template, temperature, max_tokens, context_len
     #    return gr.update(value=''), [(history[i]['content'], history[i+1]['content']) for i in range(0, len(history)-1, 2)], state
     
     #prompt_template = prompt_templates[prompt_template]
+    global prompt_template
+    prompt_template = prompt_template
     print(prompt_template)
     print(prompt_templates[prompt_template])