Update app.py

app.py

@@ -68,30 +68,20 @@ def respond(
             except json.JSONDecodeError:
                 continue  # 유효하지 않은 JSON이 있을 경우 무시하고 다음 청크로 넘어감
 
-# Gradio Blocks API 사용
-with gr.Blocks() as demo:
-    with gr.Row():
-        chatbot = gr.Chatbot()
-        with gr.Column():
-            message = gr.Textbox(label="Your message:")
-            system_message = gr.Textbox(value="AI Assistant Role", label="System message")
-            max_tokens = gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens")
-            temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
-            top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)")
-            send_button = gr.Button("Send")
+theme = "Nymbo/Nymbo_Theme"
 
-    def handle_response(message, history, system_message, max_tokens, temperature, top_p):
-        bot_response = respond(message, history, system_message, max_tokens, temperature, top_p)
-        for response in bot_response:
-            history.append((message, response))
-            yield history, history
-
-    send_button.click(
-        handle_response,
-        inputs=[message, chatbot, system_message, max_tokens, temperature, top_p],
-        outputs=[chatbot, chatbot],
-        queue=True
-    )
+# Gradio ChatInterface 설정
+demo = gr.ChatInterface(
+    fn=respond,
+    theme=theme,
+    additional_inputs=[
+        gr.Textbox(value="AI Assistant Role", label="System message"),
+        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
+        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
+        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
+    ],
+    streaming=True  # 스트리밍 모드 활성화
+)
 
 if __name__ == "__main__":
     demo.queue().launch(max_threads=20)