Update app.py

app.py

@@ -28,6 +28,15 @@ from langchain_pinecone import PineconeVectorStore
 from langchain.chains import RetrievalQA
 import asyncio
 
+from langchain.globals import set_llm_cache
+from langchain_openai import OpenAI
+from langchain.cache import SQLiteCache
+
+
+# Set up SQLite-based caching
+set_llm_cache(SQLiteCache())
+
+
 embeddings = OpenAIEmbeddings(api_key=os.environ['OPENAI_API_KEY'])
 def initialize_gpt_model():
     return ChatOpenAI(api_key=os.environ['OPENAI_API_KEY'], temperature=0, model='gpt-4o')
@@ -52,7 +61,16 @@ index_name="radarclintcountrymusic11152024"
 vectorstore = PineconeVectorStore(index_name=index_name, embedding=embeddings)
 retriever = vectorstore.as_retriever(search_kwargs={'k': 2})
 
-chat_model = ChatOpenAI(api_key=os.environ['OPENAI_API_KEY'], temperature=0, model='gpt-4o')
+#chat_model = ChatOpenAI(api_key=os.environ['OPENAI_API_KEY'], temperature=0, model='gpt-4o')
+
+# Initialize a slower LLM model with caching
+chat_model = OpenAI(
+    api_key=os.environ['OPENAI_API_KEY'],
+    model_name="gpt-4o",
+    n=2,
+    best_of=2,
+    temperature=0
+)
 
 #code for history
 conversational_memory = ConversationBufferWindowMemory(
@@ -103,17 +121,30 @@ Helpful Answer:"""
 
 QA_CHAIN_PROMPT= PromptTemplate(input_variables=["context", "question"], template=template)
 
-def build_qa_chain(prompt_template):
+#def build_qa_chain(prompt_template):
+    #qa_chain = RetrievalQA.from_chain_type(
+        #llm=chat_model,
+        #chain_type="stuff",
+        #retriever=retriever,
+        #chain_type_kwargs={"prompt": prompt_template}
+    #)
+    #return qa_chain  # Return the qa_chain object
+
+# Function to initialize QA Chain with caching enabled
+def build_qa_chain_with_cache(prompt_template):
     qa_chain = RetrievalQA.from_chain_type(
         llm=chat_model,
         chain_type="stuff",
         retriever=retriever,
         chain_type_kwargs={"prompt": prompt_template}
     )
-    return qa_chain  # Return the qa_chain object
+    return qa_chain
 
 # Instantiate the QA Chain using the defined prompt template
-qa_chain = build_qa_chain(QA_CHAIN_PROMPT)
+#qa_chain = build_qa_chain(QA_CHAIN_PROMPT)
+
+# Instantiate the QA Chain using the cached LLM
+qa_chain = build_qa_chain_with_cache(QA_CHAIN_PROMPT)
 
 # Define the function to clear input and output
 def clear_fields():