Update app.py

app.py

@@ -9,17 +9,13 @@ model_path = hf_hub_download(
     filename="AstroSage-8B-Q8_0.gguf"
 )
 
-@spaces.GPU(duration=119)
-def load_llm():
-    llm = Llama(
-        model_path=model_path,
-        n_ctx=2048,
-        chat_format="llama-3",
-        n_gpu_layers=-1,  # ensure all layers are on GPU
-    )
-    return llm
-
-llm = load_llm()
+llm = Llama(
+    model_path=model_path,
+    n_ctx=2048,
+    chat_format="llama-3",
+    n_gpu_layers=-1,  # ensure all layers are on GPU
+)
+return llm
 
 # Placeholder responses for when context is empty
 GREETING_MESSAGES = [