Update app.py

app.py

@@ -26,6 +26,7 @@ import threading
 from langchain_openai import OpenAIEmbeddings
 from langchain_pinecone import PineconeVectorStore
 from langchain.chains import RetrievalQA
+import asyncio
 
 embeddings = OpenAIEmbeddings(api_key=os.environ['OPENAI_API_KEY'])
 def initialize_gpt_model():
@@ -36,7 +37,7 @@ gpt_model = initialize_gpt_model()
 
 gpt_embeddings = OpenAIEmbeddings(api_key=os.environ['OPENAI_API_KEY'])
 gpt_vectorstore = PineconeVectorStore(index_name="radardata11122024", embedding=gpt_embeddings)
-gpt_retriever = gpt_vectorstore.as_retriever(search_kwargs={'k': 5})
+gpt_retriever = gpt_vectorstore.as_retriever(search_kwargs={'k': 2})
 
 
 
@@ -47,7 +48,7 @@ pc = Pinecone(api_key=os.environ['PINECONE_API_KEY'])
 
 index_name ="radardata11122024"
 vectorstore = PineconeVectorStore(index_name=index_name, embedding=embeddings)
-retriever = vectorstore.as_retriever(search_kwargs={'k': 5})
+retriever = vectorstore.as_retriever(search_kwargs={'k': 2})
 
 chat_model = ChatOpenAI(api_key=os.environ['OPENAI_API_KEY'], temperature=0, model='gpt-4o')
 
@@ -134,19 +135,15 @@ def generate_audio_elevenlabs(text):
 import time
 
 # Main function to handle mode selection with character-by-character streaming
-def handle_mode_selection(mode, chat_history, question):
+async def async_handle_mode_selection(mode, chat_history, question):
     if mode == "Normal Chatbot":
-        chat_history.append((question, ""))  # Append user question with an empty response initially
-
-        # Get response from Pinecone using the qa_chain
-        response = qa_chain({"query": question, "context": ""})
+        chat_history.append((question, ""))
+        response = await async_get_retriever_result(question, retriever)
         response_text = response['result']
-
-        # Stream each character in the response text to the chat history
         for i, char in enumerate(response_text):
-            chat_history[-1] = (question, chat_history[-1][1] + char)  # Update the last message
-            yield chat_history, "", None  # Yield updated chat history
-            time.sleep(0.05)  # Small delay to simulate streaming
+            chat_history[-1] = (question, chat_history[-1][1] + char)
+            yield chat_history, "", None
+            await asyncio.sleep(0.05)
 
     elif mode == "Voice to Voice Conversation":
         response_text = qa_chain({"query": question, "context": ""})['result']
@@ -188,9 +185,7 @@ def generate_audio_from_last_response(history):
             return generate_audio_elevenlabs(recent_response)
     return None
 
-# Function to insert the prompt into the textbox when clicked
-#def insert_prompt(current_text, prompt):
-    #return prompt[0] if prompt else current_text
+
 
 
 # Define the ASR model with Whisper