chore: Add GPU support

app.py

@@ -1,15 +1,17 @@
+import spaces
 import gradio as gr
 
 
 
 # Use a pipeline as a high-level helper
 from transformers import pipeline
+pipe = pipeline("text-generation", model="NousResearch/Hermes-3-Llama-3.1-8B")
 
+@spaces.GPU(duration=120)
 def llama3_1_8B(question):
     messages = [
     {"role": "user", "content": question},
     ]
-    pipe = pipeline("text-generation", model="NousResearch/Hermes-3-Llama-3.1-8B")
     responses = pipe(messages)
     return str(responses)