fix:fix some bug

app.py

@@ -6,13 +6,13 @@ import time
 import torch
 import spaces
 
-MODEL_ID = "Qwen/Qwen2-VL-2B-Instruct"
+MODEL_ID = "Qwen/Qwen2.5-VL-3B-Instruct"
 processor = AutoProcessor.from_pretrained(MODEL_ID, trust_remote_code=True)
-model = Qwen2_VLForConditionalGeneration.from_pretrained(
+model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
     MODEL_ID,
     trust_remote_code=True,
     torch_dtype=torch.bfloat16
-).to("cpu").eval()
+).to("cuda").eval()
 
 @spaces.GPU
 def model_inference(input_dict, history):
@@ -78,7 +78,7 @@ examples = [
 
 demo = gr.ChatInterface(
     fn=model_inference,
-    description="# **Qwen2-VL-2B-Instruct**",
+    description="# **Qwen2.5-VL-3B-Instruct**",
     examples=examples,
     textbox=gr.MultimodalTextbox(label="Query Input", file_types=["image"], file_count="multiple"),
     stop_btn="Stop Generation",