diff --git a/python/paddle/distributed/launch.py b/python/paddle/distributed/launch.py
index 73e91abbd4a93b230d029d776c7d80bdadeafd66..3e2f2e59f9142c36c9914abf8eabab81d3e8edbb 100644
--- a/python/paddle/distributed/launch.py
+++ b/python/paddle/distributed/launch.py
@@ -95,9 +95,9 @@ POD_IP (current node ip address, not needed for local training)
         help="The current node ip. ")
     parser.add_argument(
         "--use_paddlecloud",
-        type=bool,
-        default="False",
-        help="wheter to use paddlecloud platform to run your multi-process job.")
+        action='store_true',
+        help="wheter to use paddlecloud platform to run your multi-process job. If false, no need to set this argument."
+    )
     parser.add_argument(
         "--started_port",
         type=int,
diff --git a/python/paddle/fluid/tests/unittests/test_launch.sh b/python/paddle/fluid/tests/unittests/test_launch.sh
index 1419ba7335b247031edfc1c67eaf490db646a57b..d3b8d34e49c3848fd227cb630cedbb8e6938fead 100644
--- a/python/paddle/fluid/tests/unittests/test_launch.sh
+++ b/python/paddle/fluid/tests/unittests/test_launch.sh
@@ -11,7 +11,8 @@ export POD_IP=127.0.0.1
 export PADDLE_TRAINERS=127.0.0.1,127.0.0.2
 export PADDLE_TRAINER_ID=0
 
-distributed_args="--use_paddlecloud True --cluster_node_ips ${cluster_node_ips} --node_ip ${node_ip} --selected_gpus=0,1 --log_dir testlog"
+distributed_args="--use_paddlecloud --cluster_node_ips=${cluster_node_ips} --node_ip=${node_ip}
+--selected_gpus=0,1 --log_dir=testlog"
 python -m paddle.distributed.launch ${distributed_args} multi_process.py
 
 str1="selected_gpus:0 worker_endpoints:127.0.0.1:6170,127.0.0.1:6171,127.0.0.2:6170,127.0.0.2:6171 trainers_num:4 current_endpoint:127.0.0.1:6170 trainer_id:0"