diff --git a/demo/requirements_web_demo.txt b/demo/requirements_web_demo.txt
index 6105b71..06e529f 100644
--- a/demo/requirements_web_demo.txt
+++ b/demo/requirements_web_demo.txt
@@ -1,2 +1,2 @@
-gradio==4.31.3
+gradio==4.44.0
 modelscope-studio
\ No newline at end of file
diff --git a/demo/web_demo_audio.py b/demo/web_demo_audio.py
index 3c00a09..7637d47 100644
--- a/demo/web_demo_audio.py
+++ b/demo/web_demo_audio.py
@@ -87,7 +87,8 @@ def predict(chatbot, task_history):
     print(f"{audios=}")
     inputs = processor(text=text, audios=audios, return_tensors="pt", padding=True)
     if not _get_args().cpu_only:
-        inputs["input_ids"] = inputs.input_ids.to("cuda")
+        inputs["input_ids"] = inputs.input_ids.to(model.device)
+        inputs["attention_mask"] = inputs.attention_mask.to(model.device)
 
     generate_ids = model.generate(**inputs, max_length=256)
     generate_ids = generate_ids[:, inputs.input_ids.size(1):]