Update app.py

app.py

@@ -70,17 +70,23 @@ def chatbot_submit(message, chat_history, system_message, max_tokens_val, temper
 
     # Chama a API da NVIDIA para gerar uma resposta
     chat_history = call_nvidia_api(chat_history, max_tokens_val, temperature_val, top_p_val)
-    return chat_history
 
-# Gradio interface components
-system_msg = gr.Textbox(BASE_SYSTEM_MESSAGE, label="System Message", placeholder="System prompt.", lines=5)
-max_tokens = gr.Slider(20, 1024, label="Max Tokens", step=20, value=1024)
-temperature = gr.Slider(0.0, 1.0, label="Temperature", step=0.1, value=0.2)
-top_p = gr.Slider(0.0, 1.0, label="Top P", step=0.05, value=0.7)
+    # Extrai apenas a mensagem do assistente da resposta
+    if chat_history and chat_history[-1]["role"] == "assistant":
+        assistant_message = chat_history[-1]["content"]
+    else:
+        assistant_message = "Desculpe, ocorreu um erro ao gerar a resposta."
+
+    return assistant_message
 
 
 with gr.Blocks() as demo:
     chat_history_state = gr.State([])
+    system_msg = gr.Textbox(BASE_SYSTEM_MESSAGE, label="System Message", placeholder="System prompt.", lines=5)
+    max_tokens = gr.Slider(20, 1024, label="Max Tokens", step=20, value=1024)
+    temperature = gr.Slider(0.0, 1.0, label="Temperature", step=0.1, value=0.2)
+    top_p = gr.Slider(0.0, 1.0, label="Top P", step=0.05, value=0.7)
+
     chatbot = gr.ChatInterface(
         fn=chatbot_submit,
         additional_inputs=[system_msg, max_tokens, temperature, top_p],