decorate the right function for ZERO gpu inference

app.py

@@ -5,7 +5,6 @@ from PIL import Image
 import spaces
 from transformers import MllamaForConditionalGeneration, AutoProcessor
 
-@spaces.GPU
 # Load the Llama 3.2 Vision Model
 def load_llama_model():
  model_id = "meta-llama/Llama-3.2-11B-Vision"
@@ -21,6 +20,7 @@ def load_llama_model():
  return model, processor
 
 # Function to generate predictions for text and image
+@spaces.GPU
 def process_input(text, image=None):
  model, processor = load_llama_model()