mainfile cgpt 3

app.py

@@ -57,7 +57,29 @@ def _combine_documents(docs, document_prompt=PromptTemplate.from_template("{page
     doc_strings = [format_document(doc, document_prompt) for doc in docs]
     return document_separator.join(doc_strings)
 
+# Function to store chat history
+chistory = []
 
+def store_chat_history(role: str, content: str):
+    chistory.append({"role": role, "content": content})
+
+# Define the chain using LCEL
+def create_conversational_qa_chain(retriever, condense_llm, answer_llm):
+    condense_question_chain = RunnableLambda(
+        lambda x: {"chat_history": chistory, "question": x['question']}
+    ) | CONDENSE_QUESTION_PROMPT | condense_llm
+
+    retrieval_chain = RunnableLambda(
+        lambda x: {"standalone_question": x}
+    ) | retriever | _combine_documents
+
+    answer_chain = ANSWER_PROMPT | answer_llm
+
+    return RunnableParallel(
+        condense_question=condense_question_chain,
+        retrieve=retrieval_chain,
+        generate_answer=answer_chain
+    )
 
 # Define the Streamlit app
 def app():
@@ -67,8 +89,6 @@ def app():
             'Which retriever would you like to use?',
             ('General Medicine', 'RespiratoryFishman', 'RespiratoryMurray', 'MedMRCP2', 'OldMedicine')
         )
-
-
         
         # Define retrievers based on option
         persist_directory = {
@@ -90,18 +110,6 @@ def app():
         vectordb = Chroma(persist_directory=persist_directory, embedding_function=embedding_function, collection_name=collection_name)
         retriever = vectordb.as_retriever(search_kwargs={"k": 5})
 
-
-    # Define the chain using LCEL
-    condense_question_chain = RunnableLambda(lambda x: {"chat_history": chistory, "question": x}) | CONDENSE_QUESTION_PROMPT | llmc
-    retriever_chain = RunnableLambda(lambda x: {"standalone_question": x}) | retriever | _combine_documents
-    answer_chain = ANSWER_PROMPT | llm
-        
-    conversational_qa_chain = RunnableParallel(
-        condense_question=condense_question_chain,
-        retrieve=retriever_chain,
-        generate_answer=answer_chain
-    )    
-    
     if "messages" not in st.session_state:
         st.session_state.messages = [{"role": "assistant", "content": "How may I help you?"}]
     
@@ -119,6 +127,7 @@ def app():
             st.write(prompts2)
 
     if st.session_state.messages[-1]["role"] != "assistant":
+        conversational_qa_chain = create_conversational_qa_chain(retriever, llmc, llm)
         with st.chat_message("assistant"):
             with st.spinner("Thinking..."):
                 response = conversational_qa_chain.invoke(