Update app.py

app.py

@@ -16,7 +16,7 @@ from transformers import TextIteratorStreamer
 from threading import Thread
 import torch
 
-device = 'cuda'
+device = 'auto'
 model_id = "google/gemma-3-4b-it"
 processor = AutoProcessor.from_pretrained(model_id, use_fast=True, padding_side="left")
 model = Gemma3ForConditionalGeneration.from_pretrained(
@@ -24,6 +24,7 @@ model = Gemma3ForConditionalGeneration.from_pretrained(
     torch_dtype=torch.bfloat16
 ).to(device).eval()
 
+@torch.inference_mode()
 def process(message, history):
     """Generate the model response in streaming mode given message and history
     """