Update app.py

app.py

@@ -42,7 +42,7 @@ quantization_config = BitsAndBytesConfig(load_in_4bit=True,
 
 # Load models only once
 processor = AutoProcessor.from_pretrained(MODEL_ID)
-model = LlavaForConditionalGeneration.from_pretrained(MODEL_ID, quantization_config=quantization_config, device_map="auto").to(DEVICE)
+model = LlavaForConditionalGeneration.from_pretrained(MODEL_ID, quantization_config=quantization_config, device_map="auto")
 pipe_image_to_text = pipeline("image-to-text", model=model, model_kwargs={"quantization_config": quantization_config})
 # Initialize the text generation pipeline
 pipe_text = pipeline("text-generation", model=TEXT_MODEL_ID, model_kwargs={"quantization_config": quantization_config})