Update llavamed_inference_openvino.py

llavamed_inference_openvino.py

@@ -30,7 +30,7 @@ model_path = "llava-med-imf16-llmint4"
 # model_path = "llava-med-imint8-llmint4"
 model_name = get_model_name_from_path(model_path)
 tokenizer, model, image_processor, context_len = load_pretrained_model(
-    model_path=model_path, model_base=None, model_name=model_name, device="gpu", openvino=True, image_device="gpu"
+    model_path=model_path, model_base=None, model_name=model_name, device="gpu", openvino=True, image_device="npu"
 )
 print("loaded models")