diff --git a/auto_round/inference/backend.py b/auto_round/inference/backend.py
index 3b243bff0..7b1783998 100644
--- a/auto_round/inference/backend.py
+++ b/auto_round/inference/backend.py
@@ -582,7 +582,7 @@ def fp8_static_scheme_checker(
 
 
 BackendInfos["ipex_awq"] = BackendInfo(
-    device=["cpu"],
+    device=["xpu"],
     sym=[True, False],
     packing_format=AWQ_FORMAT,
     bits=[4],
diff --git a/test/test_cuda/test_auto_round_format.py b/test/test_cuda/test_auto_round_format.py
index 9604ffff1..821d45fa1 100644
--- a/test/test_cuda/test_auto_round_format.py
+++ b/test/test_cuda/test_auto_round_format.py
@@ -158,7 +158,7 @@ def test_autoround_gptq_sym_format(self, tiny_opt_model_path, dataloader):
 
         from transformers import AutoRoundConfig
 
-        quantization_config = AutoRoundConfig(backend="ipex_gptq")
+        quantization_config = AutoRoundConfig(backend="ipex")
 
         model = AutoModelForCausalLM.from_pretrained(
             quantized_model_path, device_map="cpu", trust_remote_code=True, quantization_config=quantization_config