Update app.py

app.py

@@ -11,6 +11,8 @@ print(f'zero.device: {zero.device}') # <-- 'cpu' 🤔
 @spaces.GPU
 def greet(n):
     global llm
+    llm = load_model(download_model())
+
     print(f'zero.device: {zero.device}') # <-- 'cuda:0' 🤗
     grammar = LlamaGrammar.from_string('''
     root ::= sentence
@@ -66,7 +68,6 @@ def load_model(fp):
     )
     return llm
 
-llm = load_model(download_model())
 print(f'llm: {llm}')
 demo = gr.Interface(fn=greet, inputs=gr.Number(), outputs=gr.Text())
 demo.launch(share=False)