Update app.py

app.py

@@ -14,6 +14,7 @@ llm = Llama(
     n_ctx=2048,
     chat_format="llama-3",
     n_gpu_layers=-1,  # ensure all layers are on GPU
+    n_threads=1, # no CPU multi-threading
 )
 
 # Placeholder responses for when context is empty