try

app.py

@@ -13,9 +13,21 @@ import subprocess
 result = subprocess.run(
     ['nvidia-smi'], text=True
 )
-##
+
+import spaces
+from threading import Thread
+from typing import Iterator
+
+import gradio as gr
 import torch
+from modelscope import AutoModelForCausalLM, AutoTokenizer
+from transformers import  TextIteratorStreamer
 
+MAX_MAX_NEW_TOKENS = 2048
+DEFAULT_MAX_NEW_TOKENS = 1024
+MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
+
+##
 # 检查CUDA是否可用
 if torch.cuda.is_available():
     print("CUDA is available. Listing available GPUs:")
@@ -31,19 +43,6 @@ else:
 
 ##
 
-import spaces
-from threading import Thread
-from typing import Iterator
-
-import gradio as gr
-import torch
-from modelscope import AutoModelForCausalLM, AutoTokenizer
-from transformers import  TextIteratorStreamer
-
-MAX_MAX_NEW_TOKENS = 2048
-DEFAULT_MAX_NEW_TOKENS = 1024
-MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
-
 
 if not torch.cuda.is_available():
     DESCRIPTION += "\n<p>Running on CPU 🥶 This demo does not work on CPU.</p>"