Updated the respond function, added chat_response and used gr.ChatInterface()

We've kept the respond function as it was, which handles the interaction with your RAG chain.

The chat_response function has been simplified. It now just calls respond and returns the bot's message.

We're using gr.Blocks() as a wrapper to include the info_text_component and the advanced options.

Inside the gr.Blocks(), we're using gr.ChatInterface() to create the chat interface. This handles most of the chat functionality for us.
We've added some optional parameters to gr.ChatInterface() like title, description, and examples to make the interface more user-friendly. You can adjust these as needed.

The advanced options (sliders) are passed as additional_inputs to the gr.ChatInterface().

app.py

@@ -112,18 +112,20 @@ info_text_component = Markdown(info_text)  # Load it once
 
 def respond(message, history, max_tokens, temperature, top_p):
     # Process user message through RAG chain
-    response = rag_chain.invoke({"input": message})# Extract the answer from the response
+    response = rag_chain.invoke({"input": message})
+    # Extract the answer from the response
     if isinstance(response, dict) and 'answer' in response:
         answer = response['answer']
     else:
         answer = str(response)  # Convert to string if it's not in the expected format
     
-    # Return the answer in the format expected by Gradio's chat interface
     return answer
 
-# Gradio Chat Interface
+def chat_response(message, history, max_tokens, temperature, top_p):
+    bot_message = respond(message, history, max_tokens, temperature, top_p)
+    return bot_message
+
 with gr.Blocks() as demo:
-    # Ensure info_text_component is only rendered once
     info_text_component.render()
 
     with gr.Accordion("Advanced Options", open=False):
@@ -131,22 +133,19 @@ with gr.Blocks() as demo:
         temperature_slider = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
         top_p_slider = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)")
 
-    chatbot = gr.Chatbot(height=600)
-    msg = gr.Textbox(
-        placeholder="Type your message here (eg. How to do great work by Paul Graham?)",
-        container=False,
-        scale=7
-    )
-
-    def chat_response(message, history, max_tokens, temperature, top_p):
-        bot_message = respond(message, history, max_tokens, temperature, top_p)
-        return bot_message
-
     gr.ChatInterface(
-        chat_response,
-        chatbot=chatbot,
-        textbox=msg,
+        fn=chat_response,
         additional_inputs=[max_tokens_slider, temperature_slider, top_p_slider],
+        title="RAG Chatbot",
+        description="Ask me anything about the documents in my knowledge base!",
+        examples=[
+            "How to do great work by Paul Graham?",
+            "What are the key points in the article about doing great work?",
+            "Can you summarize Paul Graham's advice on career success?"
+        ],
+        retry_btn="Retry",
+        undo_btn="Undo",
+        clear_btn="Clear",
     )
 
 if __name__ == "__main__":