Adding memory to chat

app.py

@@ -28,23 +28,28 @@ qa_chain = RetrievalQA.from_chain_type(
     retriever=vectordb.as_retriever()
 )
 
-question = "production is broken how do I fix it?"
-
-result = qa_chain({"query": question})
+# Conversation memory
+from langchain.memory import ConversationBufferMemory
+memory = ConversationBufferMemory(
+    memory_key="chat_history",
+    return_messages=True
+)
 
-print(result['result'])
+from langchain.chains import ConversationalRetrievalChain
+retriever=vectordb.as_retriever()
+qa_memory = ConversationalRetrievalChain.from_llm(
+    llm,
+    retriever=retriever,
+    memory=memory
+)
 
 
 # Streamed response emulator
 def response_generator(prompt):
-    # response = random.choice(
-    #     [
-    #         result['result']
-    #     ]
-    # )
-
-    response = qa_chain({"query": prompt})['result']   
-    
+
+    response = qa_memory({"query": prompt})['result']   
+
+    # Fake streaming
     for word in response.split():
         yield word + " "
         time.sleep(0.05)