Enable ZeroGPU inference

app.py

@@ -1,9 +1,14 @@
 import gradio as gr
 from transformers import pipeline
 
-pipe = pipeline("text-generation", model="Spreadsheet-RL/Spreadsheet-RL-4B")
+import spaces
 
+pipe = pipeline(
+    "text-generation", model="Spreadsheet-RL/Spreadsheet-RL-4B", device="cuda"
+)
 
+
+@spaces.GPU
 def generate(prompt):
     messages = [
         {"role": "user", "content": prompt},