Return pipe inside the function

app.py

@@ -1,23 +1,20 @@
 import spaces
 import gradio as gr
 import torch
-import ray
 
 
 # Use a pipeline as a high-level helper
 from transformers import pipeline
 
-torch.cuda.empty_cache()
 
-print("RUNNING PIPE")
-pipe = pipeline("text-generation", model="NousResearch/Hermes-3-Llama-3.1-8B", max_new_tokens=200, device=0)
-print("PIPE DONE") 
 
-@ray.remote(num_gpus=1, max_calls=1)
+
 @spaces.GPU(duration=120)
 def llama3_1_8B(question):
     
-
+    print("RUNNING PIPE")
+    pipe = pipeline("text-generation", model="NousResearch/Hermes-3-Llama-3.1-8B", max_new_tokens=200, device=0)
+    print("PIPE DONE") 
     messages = [
     {"role": "user", "content": question},
     ]