Update app.py

app.py

@@ -1,20 +1,31 @@
 import gradio as gr
 from huggingface_hub.utils import HfHubHTTPError
 
+
 def predict(message, history):
     try:
-        # Load the model and create a prediction function
-        iface = gr.Interface.load("models/meta-llama/Meta-Llama-3.1-8B")
-        fn = iface.launch(share=False,prevent_thread_lock=True)  
-        response = fn(message)  # Get the prediction using the function
+        # Create the Interface object directly
+        iface = gr.Interface(
+            fn=gr.ChatInterface.load("models/meta-llama/Meta-Llama-3.1-8B"),
+            inputs="textbox",
+            outputs="textbox",
+        )
+
+        # Launch the interface and get the prediction function
+        with iface:
+            fn = iface.predict
+        
+        response = fn(message, history)
         history.append((message, response))
         return "", history
+
     except HfHubHTTPError as e:
         if e.response.status_code == 504:
             return "Server overloaded. Please try again later.", history
         else:
             raise e
 
+
 with gr.Blocks() as demo:
     chatbot = gr.Chatbot()
     msg = gr.Textbox()