Update app.py

app.py

@@ -58,32 +58,39 @@ def generate_response(history, system_prompt):
     thread.start()
 
     # Stream tokens
+    partial_message = ""
     for token in streamer:
-        yield token
+        partial_message += token
+        yield partial_message
 
 # === Gradio interface ===
 with gr.Blocks() as demo:
     gr.Markdown("## 🧠 Chat with SBK LLM")
     system_prompt = gr.Textbox(label="System Prompt", value=SYSTEM_PROMPT, lines=2)
-    chatbot = gr.Chatbot()  # Reverted to original format
+    chatbot = gr.Chatbot()  # Using default tuple format
     msg = gr.Textbox(label="Your Message", placeholder="Ask me anything...", lines=1)
     clear = gr.Button("Clear")
 
     history = gr.State([])  # memory for session
 
-    def respond(user_message, history, system_prompt):
-        history = history + [[user_message, ""]]
-        return "", history, generate_response(history, system_prompt)
+    def respond(user_message, chat_history, system_prompt):
+        # Append user message to history
+        chat_history.append((user_message, ""))
+        
+        # Generate response
+        full_response = ""
+        for response in generate_response(chat_history, system_prompt):
+            full_response = response
+            chat_history[-1] = (user_message, full_response)
+            yield chat_history
+        
+        return "", chat_history
 
-    # Updated submit handler
     msg.submit(
         respond,
-        [msg, history, system_prompt],
-        [msg, history, chatbot]
+        [msg, chatbot, system_prompt],
+        [chatbot, history],
     )
-    clear.click(lambda: ([], "", ""), outputs=[chatbot, msg, history])
+    clear.click(lambda: ([], []), outputs=[chatbot, history])
 
-    # Enable streaming through queue
-    demo.queue()
-
-demo.launch()
+demo.queue().launch()