update

app.py

@@ -11,7 +11,7 @@ import cumo.serve.gradio_web_server as gws
 #os.system("pip install flash-attn --no-build-isolation")
 
 # Execute the pip install command with additional options
-subprocess.check_call([sys.executable, '-m', 'pip', 'install', 'flash-attn', '--no-build-isolation', '-U'])
+#subprocess.check_call([sys.executable, '-m', 'pip', 'install', 'flash-attn', '--no-build-isolation', '-U'])
 
 def start_controller():
     print("Starting the controller")
@@ -45,7 +45,6 @@ def start_worker(model_path: str, bits=16):
         model_path,
         "--model-name",
         model_name,
-        "--use-flash-attn",
     ]
     if bits != 16:
         worker_command += [f"--load-{bits}bit"]