Update app.py

app.py

@@ -8,17 +8,17 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
 
 AUTH_TOKEN = os.environ.get("HF_TOKEN", False)
 model = AutoModelForCausalLM.from_pretrained(
-    "meta-llama/Llama-3.2-1B-Instruct",
+    "meta-llama/Llama-3.2-3B-Instruct",
     token=AUTH_TOKEN,
 )
 tok = AutoTokenizer.from_pretrained(
-    "meta-llama/Llama-3.2-1B-Instruct",
+    "meta-llama/Llama-3.2-3B-Instruct",
     token=AUTH_TOKEN,
 )
 model = model.to("cuda").eval()
 
 
-@spaces.GPU
+@spaces.GPU(duration=10)
 def greet(name):
     lm = Transformers(model=model, tokenizer=tok)
     lm += "Let the user know what the temperature for today is, in farenheit.\n"