Update app.py

app.py

@@ -1,6 +1,7 @@
 import gradio as gr
 import torch
 from transformers import AutoTokenizer, AutoModelForCausalLM
+import spaces
 
 model_id = "NousResearch/Hermes-4-14B"
 
@@ -11,6 +12,7 @@ model = AutoModelForCausalLM.from_pretrained(
     device_map="auto"
 )
 
+@spaces.GPU(duration=120)
 def predict(message, history):
     history.append({"role": "user", "content": message})