diff --git a/serving/README.md b/serving/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..37d688f548a6742b11e4fbe6ab85af1daaccde66
--- /dev/null
+++ b/serving/README.md
@@ -0,0 +1 @@
+# PLSC Serving
diff --git a/serving/client/face_service/data/00000000.jpg b/serving/client/face_service/data/00000000.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..cc2c8ec8f59bade64ff622c700d8b3dbc5249c2c
Binary files /dev/null and b/serving/client/face_service/data/00000000.jpg differ
diff --git a/serving/client/face_service/face_service.py b/serving/client/face_service/face_service.py
new file mode 100644
index 0000000000000000000000000000000000000000..09c43ad260ae56d8cfbcdebdfb0c2de99aa43b97
--- /dev/null
+++ b/serving/client/face_service/face_service.py
@@ -0,0 +1,94 @@
+# coding:utf-8
+import sys
+import time
+from base64 import b64encode
+import ujson
+import os
+from PIL import Image
+
+_ver = sys.version_info
+is_py2 = (_ver[0] == 2)
+is_py3 = (_ver[0] == 3)
+
+if is_py2:
+    import httplib
+if is_py3:
+    import http.client as httplib
+
+
+class FaceService():
+    def __init__(self):
+        self.batch_size = 16
+        self.con_list = []
+        self.con_index = 0
+        self.server_list = []
+
+    def connect(self, server='127.0.0.1:8010'):
+        self.server_list.append(server)
+        con = httplib.HTTPConnection(server)
+        self.con_list.append(con)
+
+    def connect_all_server(self, server_list):
+        for server in server_list:
+            self.server_list.append(server)
+            self.con_list.append(httplib.HTTPConnection(server))
+
+    def infer(self, request_msg):
+
+        try:
+            cur_con = self.con_list[self.con_index]
+            cur_con.request('POST', "/FaceClassifyService/inference",
+                            request_msg, {"Content-Type": "application/json"})
+            response = cur_con.getresponse()
+            response_msg = response.read()
+            #print(response_msg)
+            response_msg = ujson.loads(response_msg)
+            self.con_index += 1
+            self.con_index = self.con_index % len(self.con_list)
+            return response_msg
+
+        except BaseException as err:
+            del self.con_list[self.con_index]
+            print(err)
+            if len(self.con_list) == 0:
+                print('All server failed')
+                return 'fail'
+            else:
+                self.con_index = 0
+                return 'retry'
+
+    def encode(self, images):
+        self.batch_size = len(images)
+        request = []
+        for si in range(self.batch_size):
+            request.append(b64encode(images[si]).decode('ascii'))
+
+        #request
+        request = {"base64_string": request}
+        request_msg = ujson.dumps(request)
+
+        response_msg = self.infer(request_msg)
+        result = []
+        for msg in response_msg["instance"]:
+            result.append(msg["embedding"])
+
+        #request end
+        return result
+
+    def close(self):
+        for con in self.con_list:
+            con.close()
+
+
+def test():
+    with open('./data/00000000.jpg', 'rb') as f:
+        image = f.read()
+    bc = FaceService()
+    bc.connect('127.0.0.1:8866')
+    result = bc.encode([image])
+    print(result[0])
+    bc.close()
+
+
+if __name__ == '__main__':
+    test()
diff --git a/serving/server/LICENSE b/serving/server/LICENSE
new file mode 100644
index 0000000000000000000000000000000000000000..96f1555dfe572f6dd2af1b7db9e100cd85bf9687
--- /dev/null
+++ b/serving/server/LICENSE
@@ -0,0 +1,19 @@
+Copyright (c) 2018 The Python Packaging Authority
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
diff --git a/serving/server/README.md b/serving/server/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..e6e0fcfb019a0fda74192c075da33d408299ca76
--- /dev/null
+++ b/serving/server/README.md
@@ -0,0 +1 @@
+This is a simple example package.
diff --git a/serving/server/plsc_serving/__init__.py b/serving/server/plsc_serving/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e7b8b2e4a4501833445c0006a9048fe896ec91c5
--- /dev/null
+++ b/serving/server/plsc_serving/__init__.py
@@ -0,0 +1,15 @@
+# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+__version__ = '0.1.4'
diff --git a/serving/server/plsc_serving/run/__init__.py b/serving/server/plsc_serving/run/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..54a2bdea807010c56f77fc52bb327c20dc3f2a3d
--- /dev/null
+++ b/serving/server/plsc_serving/run/__init__.py
@@ -0,0 +1,194 @@
+# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+import re
+import tarfile
+import face_serving
+import subprocess
+import imp
+import time
+
+
+class PLSCServer():
+    def __init__(self, with_gpu=True):
+        os.chdir(self.get_path())
+        self.with_gpu_flag = with_gpu
+        self.p_list = []
+        self.use_other_model = False
+        self.run_m = False
+        self.model_url = 'https://paddle-serving.bj.bcebos.com/paddle-gpu-serving/model-face'
+        self.bin_url = 'https://paddle-serving.bj.bcebos.com/paddle-gpu-serving/bin-face'
+        self.cpu_run_cmd = './bin/serving-cpu --bthread_min_concurrency=4 --bthread_concurrency=4 --logtostderr=true '
+        self.gpu_run_cmd = './bin/serving-gpu --bthread_min_concurrency=4 --bthread_concurrency=4 --logtostderr=true '
+        self.model_path_str = ''
+        self.get_exe()
+
+    def help(self):
+        print("hello")
+
+    def get_exe(self):
+        exe_path = './bin'
+        module_version = face_serving.__version__
+        target_version_list = module_version.strip().split('.')
+        target_version = target_version_list[0] + '.' + target_version_list[1]
+        need_download = False
+
+        if os.path.exists(exe_path):
+            with open('./bin/serving-version.txt') as f:
+                serving_version = f.read().strip()
+            if serving_version != target_version:
+                need_download = True
+        else:
+            need_download = True
+        if need_download:
+            tar_name = 'face-serving-' + target_version + '-bin.tar.gz'
+            bin_url = self.bin_url + '/' + tar_name
+            print('Frist time run, downloading PaddleServing components ...')
+            os.system('wget ' + bin_url + ' --no-check-certificate')
+            print('Decompressing files ..')
+            tar = tarfile.open(tar_name)
+            tar.extractall()
+            tar.close()
+            os.remove(tar_name)
+
+    def modify_conf(self, gpu_index=0):
+        os.chdir(self.get_path())
+        engine_name = 'name: "face_resnet50"'
+        if not self.with_gpu_flag:
+            with open('./conf/model_toolkit.prototxt', 'r') as f:
+                conf_str = f.read()
+            conf_str = re.sub('GPU', 'CPU', conf_str)
+            conf_str = re.sub('name.*"', engine_name, conf_str)
+            conf_str = re.sub('model_data_path.*"', self.model_path_str,
+                              conf_str)
+            conf_str = re.sub('enable_memory_optimization: 0',
+                              'enable_memory_optimization: 1', conf_str)
+            open('./conf/model_toolkit.prototxt', 'w').write(conf_str)
+
+        else:
+            conf_file = './conf/model_toolkit.prototxt.' + str(gpu_index)
+            with open(conf_file, 'r') as f:
+                conf_str = f.read()
+            conf_str = re.sub('CPU', 'GPU', conf_str)
+            conf_str = re.sub('name.*"', engine_name, conf_str)
+            conf_str = re.sub('model_data_path.*"', self.model_path_str,
+                              conf_str)
+            conf_str = re.sub('enable_memory_optimization: 0',
+                              'enable_memory_optimization: 1', conf_str)
+            open(conf_file, 'w').write(conf_str)
+
+    def hold(self):
+        try:
+            while True:
+                time.sleep(60)
+        except KeyboardInterrupt:
+            print("Server is going to quit")
+            time.sleep(5)
+
+    def run(self, gpu_index=0, port=8866):
+
+        os.chdir(self.get_path())
+        self.modify_conf(gpu_index)
+
+        if self.with_gpu_flag == True:
+            gpu_msg = '--gpuid=' + str(gpu_index) + ' '
+            run_cmd = self.gpu_run_cmd + gpu_msg
+            run_cmd += '--port=' + str(
+                port) + ' ' + '--resource_file=resource.prototxt.' + str(
+                    gpu_index) + ' '
+            print('Start serving on gpu ' + str(gpu_index) + ' port = ' + str(
+                port))
+        else:
+            re = subprocess.Popen(
+                'cat /usr/local/cuda/version.txt > tmp 2>&1', shell=True)
+            re.wait()
+            if re.returncode == 0:
+                run_cmd = self.gpu_run_cmd + '--port=' + str(port) + ' '
+            else:
+                run_cmd = self.cpu_run_cmd + '--port=' + str(port) + ' '
+            print('Start serving on cpu port = {}'.format(port))
+
+        process = subprocess.Popen(run_cmd, shell=True)
+
+        self.p_list.append(process)
+        if not self.run_m:
+            self.hold()
+
+    def run_multi(self, gpu_index_list=[], port_list=[]):
+        self.run_m = True
+        if len(port_list) < 1:
+            print('Please set one port at least.')
+            return -1
+        if self.with_gpu_flag == True:
+            if len(gpu_index_list) != len(port_list):
+                print('Expect same length of gpu_index_list and port_list.')
+                return -1
+            for gpu_index, port in zip(gpu_index_list, port_list):
+                self.run(gpu_index=gpu_index, port=port)
+        else:
+            for port in port_list:
+                self.run(port=port)
+        self.hold()
+
+    def stop(self):
+        for p in self.p_list:
+            p.kill()
+
+    def show_conf(self):
+        '''
+        with open('./conf/model_toolkit.prototxt', 'r') as f:
+            conf_str = f.read()
+        print(conf_str)
+        '''
+
+    def with_model(self, model_name=None, model_url=None):
+        '''
+        if model_url != None:
+            self.mode_url = model_url
+            self.use_other_model = True
+        '''
+        if model_name == None or type(model_name) != str:
+            print('Please set model name string')
+        os.chdir(self.get_path())
+        self.get_model(model_name)
+
+    def get_path(self):
+        py_path = os.path.dirname(face_serving.__file__)
+        server_path = os.path.join(py_path, 'server')
+        return server_path
+
+    def get_model(self, model_name):
+        server_path = self.get_path()
+        if True:
+            tar_name = model_name + '.tar.gz'
+            model_url = self.model_url + '/' + tar_name
+
+            model_path = os.path.join(server_path, 'data/model/paddle/fluid')
+            if not os.path.exists(model_path):
+                os.makedirs('data/model/paddle/fluid')
+            os.chdir(model_path)
+            if os.path.exists(model_name):
+                pass
+            else:
+                os.system('wget ' + model_url + ' --no-check-certificate')
+                print('Decompressing files ..')
+                tar = tarfile.open(tar_name)
+                tar.extractall()
+                tar.close()
+                os.remove(tar_name)
+
+            self.model_path_str = r'model_data_path: "./data/model/paddle/fluid/' + model_name + r'"'
+
+        os.chdir(self.get_path())
diff --git a/serving/server/plsc_serving/server/conf/cube.conf b/serving/server/plsc_serving/server/conf/cube.conf
new file mode 100644
index 0000000000000000000000000000000000000000..35310301f606a2502ea5a233f480e7656fa7583e
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/cube.conf
@@ -0,0 +1,15 @@
+[{
+    "dict_name": "dict",
+    "shard": 2,
+    "dup": 1,
+    "timeout": 200,
+    "retry": 3,
+    "backup_request": 100,
+    "type": "ipport_list",
+    "load_balancer": "rr",
+    "nodes": [{
+        "ipport_list": "list://xxx.xxx.xxx.xxx:8000"
+    },{ 
+        "ipport_list": "list://xxx.xxx.xxx.xxx:8000"
+    }]  
+}]
diff --git a/serving/server/plsc_serving/server/conf/gflags.conf b/serving/server/plsc_serving/server/conf/gflags.conf
new file mode 100644
index 0000000000000000000000000000000000000000..f9c735b5e272f528eeaa409762c9c25bba09dcf3
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/gflags.conf
@@ -0,0 +1,2 @@
+--enable_model_toolkit
+--enable_cube=false
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt
new file mode 100644
index 0000000000000000000000000000000000000000..a3ff5b5c87f2da2eab47e6f205ffaf3115335d64
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt
@@ -0,0 +1,11 @@
+engines {
+  name: "face_classify_model"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.0 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.0
new file mode 100644
index 0000000000000000000000000000000000000000..7797344772fef223743c849c84cfab4985f5e900
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.0
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 0
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.1 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.1
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.1
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.10 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.10
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.10
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.11 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.11
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.11
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.12 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.12
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.12
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.13 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.13
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.13
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.14 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.14
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.14
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.15 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.15
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.15
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.2 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.2
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.2
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.3 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.3
new file mode 100644
index 0000000000000000000000000000000000000000..7797344772fef223743c849c84cfab4985f5e900
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.3
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 0
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.4 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.4
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.4
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.5 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.5
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.5
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.6 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.6
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.6
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.7 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.7
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.7
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.8 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.8
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.8
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.9 b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.9
new file mode 100644
index 0000000000000000000000000000000000000000..a5352b790dc2b4e3e8628d22da2196948974336a
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/model_toolkit.prototxt.9
@@ -0,0 +1,11 @@
+engines {
+  name: "bert"
+  type: "FLUID_GPU_ANALYSIS_DIR"
+  reloadable_meta: "./data/model/paddle/fluid_time_file"
+  reloadable_type: "timestamp_ne"
+  model_data_path: "/home/xulongteng/.paddlehub/bert_service/bert_chinese_L-12_H-768_A-12"
+  runtime_thread_num: 0
+  batch_infer_size: 0
+  enable_batch_align: 0
+  enable_memory_optimization: 1
+}
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt b/serving/server/plsc_serving/server/conf/resource.prototxt
new file mode 100644
index 0000000000000000000000000000000000000000..0a0d6678fe99a89517ba9855140c0c01f919742e
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.0 b/serving/server/plsc_serving/server/conf/resource.prototxt.0
new file mode 100644
index 0000000000000000000000000000000000000000..c202367753ae63f95918ade7d8c10098f974f189
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.0
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.0"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.1 b/serving/server/plsc_serving/server/conf/resource.prototxt.1
new file mode 100644
index 0000000000000000000000000000000000000000..d7c15d4db80938c424dd8749a217332b4f2c83a2
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.1
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.1"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.10 b/serving/server/plsc_serving/server/conf/resource.prototxt.10
new file mode 100644
index 0000000000000000000000000000000000000000..74b0033a4a83985656de5f43af05d10faa8f2e33
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.10
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.10"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.11 b/serving/server/plsc_serving/server/conf/resource.prototxt.11
new file mode 100644
index 0000000000000000000000000000000000000000..91bd7417debd0c95c65de4bbe39d55d003a33356
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.11
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.11"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.12 b/serving/server/plsc_serving/server/conf/resource.prototxt.12
new file mode 100644
index 0000000000000000000000000000000000000000..483b713d787f75b5699e6a18281fb958c50d6348
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.12
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.12"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.13 b/serving/server/plsc_serving/server/conf/resource.prototxt.13
new file mode 100644
index 0000000000000000000000000000000000000000..0a3d25a045d16e5cd005215a5f62961de515bcc7
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.13
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.13"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.14 b/serving/server/plsc_serving/server/conf/resource.prototxt.14
new file mode 100644
index 0000000000000000000000000000000000000000..64cb4cffe879d2ecd33ecf943964b57ba86b6a98
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.14
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.14"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.15 b/serving/server/plsc_serving/server/conf/resource.prototxt.15
new file mode 100644
index 0000000000000000000000000000000000000000..462193790116d1bb39e56e3179d1e9236c2fc9ce
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.15
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.15"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.2 b/serving/server/plsc_serving/server/conf/resource.prototxt.2
new file mode 100644
index 0000000000000000000000000000000000000000..45509468a088ec3c3375b2a245a36ab42d83403b
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.2
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.2"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.3 b/serving/server/plsc_serving/server/conf/resource.prototxt.3
new file mode 100644
index 0000000000000000000000000000000000000000..3aadbe34c60c8a3b36199f166847b307b1831c1d
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.3
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.3"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.4 b/serving/server/plsc_serving/server/conf/resource.prototxt.4
new file mode 100644
index 0000000000000000000000000000000000000000..eec742fc0e4dae052ff0d656e105d9a238d02cf9
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.4
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.4"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.5 b/serving/server/plsc_serving/server/conf/resource.prototxt.5
new file mode 100644
index 0000000000000000000000000000000000000000..5db39ee97ad93a47e9713d1cfdd91b0f5bf49b3c
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.5
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.5"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.6 b/serving/server/plsc_serving/server/conf/resource.prototxt.6
new file mode 100644
index 0000000000000000000000000000000000000000..ae65b2516959782282d6fc1247edbf743ba84235
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.6
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.6"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.7 b/serving/server/plsc_serving/server/conf/resource.prototxt.7
new file mode 100644
index 0000000000000000000000000000000000000000..592f0f15cf3096be8bc7abb0207831ded5c64dda
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.7
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.7"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.8 b/serving/server/plsc_serving/server/conf/resource.prototxt.8
new file mode 100644
index 0000000000000000000000000000000000000000..ccd8497170dcc679d7343a2b95e4fb005b950bd7
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.8
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.8"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/resource.prototxt.9 b/serving/server/plsc_serving/server/conf/resource.prototxt.9
new file mode 100644
index 0000000000000000000000000000000000000000..5e1238110ae83e6a5d99ba1037e4e8112bf126b8
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/resource.prototxt.9
@@ -0,0 +1,3 @@
+model_toolkit_path: "./conf/"
+model_toolkit_file: "model_toolkit.prototxt.9"
+cube_config_file: "./conf/cube.conf"
diff --git a/serving/server/plsc_serving/server/conf/service.prototxt b/serving/server/plsc_serving/server/conf/service.prototxt
new file mode 100644
index 0000000000000000000000000000000000000000..7774d40f8f18a5212b4ad26cf98551bddd53c9ce
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/service.prototxt
@@ -0,0 +1,4 @@
+services {
+  name: "FaceClassifyService"
+  workflows: "workflow10"
+}
diff --git a/serving/server/plsc_serving/server/conf/workflow.prototxt b/serving/server/plsc_serving/server/conf/workflow.prototxt
new file mode 100644
index 0000000000000000000000000000000000000000..56daa99284a07d490a3782e3d3849f90ae71bee1
--- /dev/null
+++ b/serving/server/plsc_serving/server/conf/workflow.prototxt
@@ -0,0 +1,8 @@
+workflows {
+  name: "workflow10"
+  workflow_type: "Sequence"
+  nodes {
+    name: "face_classify_op"
+    type: "FaceClassifyOp"
+  }
+}
diff --git a/serving/server/plsc_serving/server/data/model/paddle/fluid_reload_flag b/serving/server/plsc_serving/server/data/model/paddle/fluid_reload_flag
new file mode 100644
index 0000000000000000000000000000000000000000..a1866984b04a26a09bbfc4ef4f08dec5e38f818d
--- /dev/null
+++ b/serving/server/plsc_serving/server/data/model/paddle/fluid_reload_flag
@@ -0,0 +1,2 @@
+paddle fluid model
+time:20180531
diff --git a/serving/server/plsc_serving/server/data/model/paddle/fluid_time_file b/serving/server/plsc_serving/server/data/model/paddle/fluid_time_file
new file mode 100644
index 0000000000000000000000000000000000000000..4d9422cd37f6c51800055f1b1dc625e2c455fc6d
--- /dev/null
+++ b/serving/server/plsc_serving/server/data/model/paddle/fluid_time_file
@@ -0,0 +1,2 @@
+201805311000
+model paddle fluid
diff --git a/serving/server/setup.py b/serving/server/setup.py
new file mode 100644
index 0000000000000000000000000000000000000000..8b84f844aa7787a0526a82662faadc4ce4a58a2c
--- /dev/null
+++ b/serving/server/setup.py
@@ -0,0 +1,50 @@
+# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import setuptools
+
+with open("README.md", "r") as fh:
+    long_description = fh.read()
+
+#read info
+info_py = './plsc_serving/__init__.py'
+info_content = open(info_py, 'r').readlines()
+version_line = [
+    l.strip() for l in info_content if l.startswith('__version__')
+][0]
+exec(version_line)  # produce __version__
+
+setuptools.setup(
+    name="plsc-serving",
+    version=__version__,
+    author="MRXLT",
+    author_email="xlt2024@gmail.com",
+    description="package for plsc serving",
+    long_description=long_description,
+    long_description_content_type="text/markdown",
+    url="https://github.com/PaddlePaddle/PLSC",
+    packages=setuptools.find_packages(),
+    classifiers=[
+        "Programming Language :: Python :: 3",
+        "License :: OSI Approved :: MIT License",
+        "Operating System :: OS Independent",
+    ],
+    python_requires='>=3.5',
+    package_data={
+        'plsc_serving': [
+            'server/conf/*',
+            'server/data/model/paddle/fluid_reload_flag',
+            'server/data/model/paddle/fluid_time_file',
+        ]
+    })