diff --git a/demo/nas/block_sa_nas_mobilenetv2.py b/demo/nas/block_sa_nas_mobilenetv2.py
index 075c7068b1dc263e3d09592e115c0d1fdb8f2882..b58fc9effee2f8d5675edb780a45a8d263804c98 100644
--- a/demo/nas/block_sa_nas_mobilenetv2.py
+++ b/demo/nas/block_sa_nas_mobilenetv2.py
@@ -20,7 +20,7 @@ reduce_rate = 0.85
 init_temperature = 10.24
 max_flops = 321208544
 server_address = ""
-port = 8909
+port = 8979
 retain_epoch = 5
 
 
@@ -243,42 +243,7 @@ if __name__ == '__main__':
         type=int,
         default=100,
         help='controller server number.')
-    # optimizer args
-    parser.add_argument(
-        '--lr_strategy',
-        type=str,
-        default='piecewise_decay',
-        help='learning rate decay strategy.')
     parser.add_argument('--lr', type=float, default=0.1, help='learning rate.')
-    parser.add_argument(
-        '--l2_decay', type=float, default=1e-4, help='learning rate decay.')
-    parser.add_argument(
-        '--step_epochs',
-        nargs='+',
-        type=int,
-        default=[30, 60, 90],
-        help="piecewise decay step")
-    parser.add_argument(
-        '--momentum_rate',
-        type=float,
-        default=0.9,
-        help='learning rate decay.')
-    parser.add_argument(
-        '--warm_up_epochs',
-        type=float,
-        default=5.0,
-        help='learning rate decay.')
-    parser.add_argument(
-        '--num_epochs', type=int, default=120, help='learning rate decay.')
-    parser.add_argument(
-        '--decay_epochs', type=float, default=2.4, help='learning rate decay.')
-    parser.add_argument(
-        '--decay_rate', type=float, default=0.97, help='learning rate decay.')
-    parser.add_argument(
-        '--total_images',
-        type=int,
-        default=1281167,
-        help='learning rate decay.')
     args = parser.parse_args()
     print(args)
 
diff --git a/demo/nas/sa_nas_mobilenetv2.py b/demo/nas/sa_nas_mobilenetv2.py
index 990a9629d8c53bc9d112fb1e69a609c239054852..5e60da0fec30210afc177f0cdbe65b5154b73472 100644
--- a/demo/nas/sa_nas_mobilenetv2.py
+++ b/demo/nas/sa_nas_mobilenetv2.py
@@ -22,7 +22,7 @@ reduce_rate = 0.85
 init_temperature = 10.24
 max_flops = 321208544
 server_address = ""
-port = 8909
+port = 8989
 retain_epoch = 5
 
 
@@ -284,6 +284,8 @@ if __name__ == '__main__':
         help='Whether to use GPU in train/test model.')
     parser.add_argument(
         '--batch_size', type=int, default=256, help='batch size.')
+    parser.add_argument(
+        '--class_dim', type=int, default=1000, help='classify number.')
     parser.add_argument(
         '--data',
         type=str,
@@ -300,43 +302,7 @@ if __name__ == '__main__':
         type=int,
         default=100,
         help='controller server number.')
-    parser.add_argument(
-        '--lr_strategy',
-        type=str,
-        default='cosine_decay',
-        help='learning rate decay strategy.')
     parser.add_argument('--lr', type=float, default=0.1, help='learning rate.')
-    parser.add_argument(
-        '--l2_decay', type=float, default=1e-4, help='learning rate decay.')
-    parser.add_argument(
-        '--class_dim', type=int, default=100, help='classify number.')
-    parser.add_argument(
-        '--step_epochs',
-        nargs='+',
-        type=int,
-        default=[30, 60, 90],
-        help="piecewise decay step")
-    parser.add_argument(
-        '--momentum_rate',
-        type=float,
-        default=0.9,
-        help='learning rate decay.')
-    parser.add_argument(
-        '--warm_up_epochs',
-        type=float,
-        default=5.0,
-        help='learning rate decay.')
-    parser.add_argument(
-        '--num_epochs', type=int, default=120, help='learning rate decay.')
-    parser.add_argument(
-        '--decay_epochs', type=float, default=2.4, help='learning rate decay.')
-    parser.add_argument(
-        '--decay_rate', type=float, default=0.97, help='learning rate decay.')
-    parser.add_argument(
-        '--total_images',
-        type=int,
-        default=1281167,
-        help='learning rate decay.')
     args = parser.parse_args()
     print(args)
 
diff --git a/demo/optimizer.py b/demo/optimizer.py
index 73f441f897d22c10d2d6e05afaa7491b227b27d4..0f0c57985f839097e9e1ae4643ba2e5a2fb64698 100644
--- a/demo/optimizer.py
+++ b/demo/optimizer.py
@@ -23,6 +23,16 @@ import paddle.fluid.layers.ops as ops
 from paddle.fluid.initializer import init_on_cpu
 from paddle.fluid.layers.learning_rate_scheduler import _decay_step_counter
 
+lr_strategy = 'cosine_decay'
+l2_decay = 1e-4
+step_epochs = [30, 60, 90]
+momentum_rate = 0.9
+warm_up_epochs = 5.0
+num_epochs = 120
+decay_epochs = 2.4
+decay_rate = 0.97
+total_images = 1281167
+
 
 def cosine_decay(learning_rate, step_each_epoch, epochs=120):
     """Applies cosine decay to the learning rate.
@@ -152,15 +162,15 @@ class Optimizer(object):
     def __init__(self, args):
         self.batch_size = args.batch_size
         self.lr = args.lr
-        self.lr_strategy = args.lr_strategy
-        self.l2_decay = args.l2_decay
-        self.momentum_rate = args.momentum_rate
-        self.step_epochs = args.step_epochs
-        self.num_epochs = args.num_epochs
-        self.warm_up_epochs = args.warm_up_epochs
-        self.decay_epochs = args.decay_epochs
-        self.decay_rate = args.decay_rate
-        self.total_images = args.total_images
+        self.lr_strategy = lr_strategy
+        self.l2_decay = l2_decay
+        self.momentum_rate = momentum_rate
+        self.step_epochs = step_epochs
+        self.num_epochs = num_epochs
+        self.warm_up_epochs = warm_up_epochs
+        self.decay_epochs = decay_epochs
+        self.decay_rate = decay_rate
+        self.total_images = total_images
 
         self.step = int(math.ceil(float(self.total_images) / self.batch_size))
 
@@ -295,6 +305,6 @@ class Optimizer(object):
 
 def create_optimizer(args):
     Opt = Optimizer(args)
-    optimizer = getattr(Opt, args.lr_strategy)()
+    optimizer = getattr(Opt, lr_strategy)()
 
     return optimizer