Merge branch 'v0.6.0' of https://github.com/PaddlePaddle/Serving into v0.6.0

doc/LOW_PRECISION_DEPLOYMENT.md

@@ -17,7 +17,7 @@ python -m paddle_serving_client.convert --dirname ResNet50_quant
 ```
 Start RPC service, specify the GPU id and precision mode
 ```
-python -m paddle_serving_server.serve --model serving_server --port 9393 --gpu_ids 0 --use_gpu --use_trt --precision int8 
+python -m paddle_serving_server.serve --model serving_server --port 9393 --gpu_ids 0 --use_trt --precision int8 
 ```
 Request the serving service with Client
 ```
@@ -27,7 +27,7 @@ from paddle_serving_app.reader import RGB2BGR, Transpose, Div, Normalize
 
 client = Client()
 client.load_client_config(
-    "resnet_v2_50_imagenet_client/serving_client_conf.prototxt")
+    "serving_client/serving_client_conf.prototxt")
 client.connect(["127.0.0.1:9393"])
 
 seq = Sequential([
@@ -37,11 +37,11 @@ seq = Sequential([
 
 image_file = "daisy.jpg"
 img = seq(image_file)
-fetch_map = client.predict(feed={"image": img}, fetch=["score"])
-print(fetch_map["score"].reshape(-1))
+fetch_map = client.predict(feed={"image": img}, fetch=["save_infer_model/scale_0.tmp_0"])
+print(fetch_map["save_infer_model/scale_0.tmp_0"].reshape(-1))
 ```
 
 ## Reference
 * [PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim)
 * [Deploy the quantized model Using Paddle Inference on Intel CPU](https://paddle-inference.readthedocs.io/en/latest/optimize/paddle_x86_cpu_int8.html)
-* [Deploy the quantized model Using Paddle Inference on Nvidia GPU](https://paddle-inference.readthedocs.io/en/latest/optimize/paddle_trt.html)
+* [Deploy the quantized model Using Paddle Inference on Nvidia GPU](https://paddle-inference.readthedocs.io/en/latest/optimize/paddle_trt.html)
\ No newline at end of file

doc/LOW_PRECISION_DEPLOYMENT_CN.md

@@ -16,7 +16,7 @@ python -m paddle_serving_client.convert --dirname ResNet50_quant
 ```
 启动rpc服务, 设定所选GPU id、部署模型精度
 ```
-python -m paddle_serving_server.serve --model serving_server --port 9393 --gpu_ids 0 --use_gpu --use_trt --precision int8 
+python -m paddle_serving_server.serve --model serving_server --port 9393 --gpu_ids 0 --use_trt --precision int8 
 ```
 使用client进行请求
 ```
@@ -43,4 +43,4 @@ print(fetch_map["score"].reshape(-1))
 ## 参考文档
 * [PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim)
 * PaddleInference Intel CPU部署量化模型[文档](https://paddle-inference.readthedocs.io/en/latest/optimize/paddle_x86_cpu_int8.html)
-* PaddleInference NV GPU部署量化模型[文档](https://paddle-inference.readthedocs.io/en/latest/optimize/paddle_trt.html)
+* PaddleInference NV GPU部署量化模型[文档](https://paddle-inference.readthedocs.io/en/latest/optimize/paddle_trt.html)
\ No newline at end of file

python/examples/low_precision/resnet50/README.md

+# resnet50 int8 example
+(English|[简体中文](./README_CN.md))
+
+## Obtain the quantized model through PaddleSlim tool
+Train the low-precision models please refer to [PaddleSlim](https://paddleslim.readthedocs.io/zh_CN/latest/tutorials/quant/overview.html).
+
+## Deploy the quantized model from PaddleSlim using Paddle Serving with Nvidia TensorRT int8 mode
+
+Firstly, download the [Resnet50 int8 model](https://paddle-inference-dist.bj.bcebos.com/inference_demo/python/resnet50/ResNet50_quant.tar.gz) and convert to Paddle Serving's saved model。
+```
+wget https://paddle-inference-dist.bj.bcebos.com/inference_demo/python/resnet50/ResNet50_quant.tar.gz
+tar zxvf ResNet50_quant.tar.gz
+
+python -m paddle_serving_client.convert --dirname ResNet50_quant
+```
+Start RPC service, specify the GPU id and precision mode
+```
+python -m paddle_serving_server.serve --model serving_server --port 9393 --gpu_ids 0 --use_trt --precision int8 
+```
+Request the serving service with Client
+```
+python resnet50_client.py
+```
+
+## Reference
+* [PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim)
+* [Deploy the quantized model Using Paddle Inference on Intel CPU](https://paddle-inference.readthedocs.io/en/latest/optimize/paddle_x86_cpu_int8.html)
+* [Deploy the quantized model Using Paddle Inference on Nvidia GPU](https://paddle-inference.readthedocs.io/en/latest/optimize/paddle_trt.html)

python/examples/low_precision/resnet50/README_CN.md

+# resnet50 int8示例
+(简体中文|[English](./README.md))
+
+## 通过PaddleSlim量化生成低精度模型
+详细见[PaddleSlim量化](https://paddleslim.readthedocs.io/zh_CN/latest/tutorials/quant/overview.html)
+
+## 使用TensorRT int8加载PaddleSlim Int8量化模型进行部署
+首先下载Resnet50 [PaddleSlim量化模型](https://paddle-inference-dist.bj.bcebos.com/inference_demo/python/resnet50/ResNet50_quant.tar.gz),并转换为Paddle Serving支持的部署模型格式。
+```
+wget https://paddle-inference-dist.bj.bcebos.com/inference_demo/python/resnet50/ResNet50_quant.tar.gz
+tar zxvf ResNet50_quant.tar.gz
+
+python -m paddle_serving_client.convert --dirname ResNet50_quant
+```
+启动rpc服务, 设定所选GPU id、部署模型精度
+```
+python -m paddle_serving_server.serve --model serving_server --port 9393 --gpu_ids 0 --use_trt --precision int8 
+```
+使用client进行请求
+```
+python resnet50_client.py
+```
+
+## 参考文档
+* [PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim)
+* PaddleInference Intel CPU部署量化模型[文档](https://paddle-inference.readthedocs.io/en/latest/optimize/paddle_x86_cpu_int8.html)
+* PaddleInference NV GPU部署量化模型[文档](https://paddle-inference.readthedocs.io/en/latest/optimize/paddle_trt.html)

python/examples/xpu/fit_a_line_xpu/test_multi_process_client.py → python/examples/low_precision/resnet50/resnet50_client.py

@@ -13,30 +13,20 @@
 # limitations under the License.
 
 from paddle_serving_client import Client
-from paddle_serving_client.utils import MultiThreadRunner
-import paddle
-import numpy as np
+from paddle_serving_app.reader import Sequential, File2Image, Resize, CenterCrop
+from paddle_serving_app.reader import RGB2BGR, Transpose, Div, Normalize
 
+client = Client()
+client.load_client_config(
+    "serving_client/serving_client_conf.prototxt")
+client.connect(["127.0.0.1:9393"])
 
-def single_func(idx, resource):
-    client = Client()
-    client.load_client_config(
-        "./uci_housing_client/serving_client_conf.prototxt")
-    client.connect(["127.0.0.1:9293", "127.0.0.1:9292"])
-    x = [
-        0.0137, -0.1136, 0.2553, -0.0692, 0.0582, -0.0727, -0.1583, -0.0584,
-        0.6283, 0.4919, 0.1856, 0.0795, -0.0332
-    ]
-    x = np.array(x)
-    for i in range(1000):
-        fetch_map = client.predict(feed={"x": x}, fetch=["price"])
-        if fetch_map is None:
-            return [[None]]
-    return [[0]]
+seq = Sequential([
+    File2Image(), Resize(256), CenterCrop(224), RGB2BGR(), Transpose((2, 0, 1)),
+    Div(255), Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225], True)
+])
 
-
-multi_thread_runner = MultiThreadRunner()
-thread_num = 4
-result = multi_thread_runner.run(single_func, thread_num, {})
-if None in result[0]:
-    exit(1)
+image_file = "daisy.jpg"
+img = seq(image_file)
+fetch_map = client.predict(feed={"image": img}, fetch=["save_infer_model/scale_0.tmp_0"])
+print(fetch_map["save_infer_model/scale_0.tmp_0"].reshape(-1))

python/examples/xpu/bert/README.md

 
 ## Prepare
+### download model and extract
+```
+wget https://paddle-inference-dist.cdn.bcebos.com/PaddleLite/models_and_data_for_unittests/bert_base_chinese.tar.gz
+tar zxvf bert_base_chinese.tar.gz
+```
 ### convert model
 ```
-python -m paddle_serving_client.convert --dirname infer_bert-base-chinese_ft_model_4000.pdparams
+python3 -m paddle_serving_client.convert --dirname bert_base_chinese --model_filename bert_base_chinese/model.pdmodel --params_filename bert_base_chinese/model.pdiparams
+```
+### or, you can get the serving saved model directly
+```
+wget https://paddle-serving.bj.bcebos.com/models/xpu/bert.tar.gz
+tar zxvf bert.tar.gz 
+```
+### Getting Dict and Sample Dataset
+
+```
+sh get_data.sh
 ```
+this script will download Chinese Dictionary File vocab.txt and Chinese Sample Data data-c.txt
 
 ## RPC Service
 
 ### Start Service
 
 ```
-pytyon bert_web_service.py serving_server 7703
+python3 bert_web_service.py serving_server 7703
 ```
 
 ### Client Prediction
 
 ```
-python bert_client.py
+head data-c.txt | python3 bert_client.py --model serving_client/serving_client_conf.prototxt
 ```

python/examples/xpu/bert/get_data.sh

+wget https://paddle-serving.bj.bcebos.com/bert_example/data-c.txt --no-check-certificate
+wget https://paddle-serving.bj.bcebos.com/bert_example/vocab.txt --no-check-certificate

python/examples/xpu/ernie/README.md

 
 ## Prepare
+### download model and extract
+```
+wget https://paddle-inference-dist.cdn.bcebos.com/PaddleLite/models_and_data_for_unittests/ernie.tar.gz
+tar zxvf ernie.tar.gz
+```
 ### convert model
 ```
-python3 -m paddle_serving_client.convert --dirname erine
+python3 -m paddle_serving_client.convert --dirname ernie
+```
+### or, you can get the serving saved model directly
+```
+wget https://paddle-serving.bj.bcebos.com/models/xpu/ernie.tar.gz
+tar zxvf ernie.tar.gz 
+```
+### Getting Dict and Sample Dataset
+
+```
+sh get_data.sh
 ```
+this script will download Chinese Dictionary File vocab.txt and Chinese Sample Data data-c.txt
 
 ## RPC Service
 

python/examples/xpu/ernie/ernie_client.py

@@ -23,7 +23,7 @@ args = benchmark_args()
 
 reader = ChineseErnieReader({"max_seq_len": 128})
 fetch = ["save_infer_model/scale_0"]
-endpoint_list = ['127.0.0.1:7704']
+endpoint_list = ['127.0.0.1:12000']
 client = Client()
 client.load_client_config(args.model)
 client.connect(endpoint_list)

python/examples/xpu/ernie/get_data.sh

+wget https://paddle-serving.bj.bcebos.com/bert_example/data-c.txt --no-check-certificate
+wget https://paddle-serving.bj.bcebos.com/bert_example/vocab.txt --no-check-certificate

python/examples/xpu/fit_a_line_xpu/README.md

@@ -8,15 +8,10 @@
 sh get_data.sh
 ```
 
-
-
 ## RPC service
 
 ### Start server
-``` shell
-python test_server.py uci_housing_model/
-```
-You can alse use the following code to start the RPC service 
+You can use the following code to start the RPC service 
 ```shell
 python -m paddle_serving_server.serve --model uci_housing_model --thread 10 --port 9393 --use_lite --use_xpu --ir_optim
 ```
@@ -29,19 +24,17 @@ The `paddlepaddle` package is used in `test_client.py`, and you may need to down
 python test_client.py uci_housing_client/serving_client_conf.prototxt
 ```
 
-
-
 ## HTTP service
 
 ### Start server
 
 Start a web service with default web service hosting modules:
 ``` shell
-python test_server.py
+python -m paddle_serving_server.serve --model uci_housing_model --thread 10 --port 9393 --use_lite --use_xpu --ir_optim --name uci
 ```
 
 ### Client prediction
 
 ``` shell
-curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"x": [0.0137, -0.1136, 0.2553, -0.0692, 0.0582, -0.0727, -0.1583, -0.0584, 0.6283, 0.4919, 0.1856, 0.0795, -0.0332]}], "fetch":["price"]}' http://127.0.0.1:9292/uci/prediction
+curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"x": [0.0137, -0.1136, 0.2553, -0.0692, 0.0582, -0.0727, -0.1583, -0.0584, 0.6283, 0.4919, 0.1856, 0.0795, -0.0332]}], "fetch":["price"]}' http://127.0.0.1:9393/uci/prediction
 ```

python/examples/xpu/fit_a_line_xpu/README_CN.md

@@ -32,8 +32,6 @@ python -m paddle_serving_server.serve --model uci_housing_model --thread 10 --po
 python test_client.py uci_housing_client/serving_client_conf.prototxt
 ```
 
-
-
 ## HTTP服务
 
 ### 开启服务端
@@ -41,11 +39,11 @@ python test_client.py uci_housing_client/serving_client_conf.prototxt
 通过下面的一行代码开启默认web服务：
 
 ``` shell
-python test_server.py
+python -m paddle_serving_server.serve --model uci_housing_model --thread 10 --port 9393 --use_lite --use_xpu --ir_optim --name uci
 ```
 
 ### 客户端预测
 
 ``` shell
-curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"x": [0.0137, -0.1136, 0.2553, -0.0692, 0.0582, -0.0727, -0.1583, -0.0584, 0.6283, 0.4919, 0.1856, 0.0795, -0.0332]}], "fetch":["price"]}' http://127.0.0.1:9292/uci/prediction
+curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"x": [0.0137, -0.1136, 0.2553, -0.0692, 0.0582, -0.0727, -0.1583, -0.0584, 0.6283, 0.4919, 0.1856, 0.0795, -0.0332]}], "fetch":["price"]}' http://127.0.0.1:9393/uci/prediction
 ```

python/examples/xpu/fit_a_line_xpu/test_server.py

-# Copyright (c) 2020 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.
-# pylint: disable=doc-string-missing
-
-from paddle_serving_server.web_service import WebService
-import numpy as np
-
-
-class UciService(WebService):
-    def preprocess(self, feed=[], fetch=[]):
-        feed_batch = []
-        is_batch = True
-        new_data = np.zeros((len(feed), 1, 13)).astype("float32")
-        for i, ins in enumerate(feed):
-            nums = np.array(ins["x"]).reshape(1, 1, 13)
-            new_data[i] = nums
-        feed = {"x": new_data}
-        return feed, fetch, is_batch
-
-
-uci_service = UciService(name="uci")
-uci_service.load_model_config("uci_housing_model")
-uci_service.prepare_server(
-    workdir="workdir", port=9393, use_lite=True, use_xpu=True, ir_optim=True)
-uci_service.run_rpc_service()
-uci_service.run_web_service()

python/examples/xpu/vgg19/README.md

 
 ## Prepare
+### download model and extract
+```
+wget https://paddle-inference-dist.bj.bcebos.com/PaddleLite/models_and_data_for_unittests/VGG19.tar.gz
+tar zxvf VGG19.tar.gz
+```
 ### convert model
 ```
-python -m paddle_serving_client.convert --dirname VGG19
+python3 -m paddle_serving_client.convert --dirname VGG19
+```
+### or, you can get the serving saved model directly
+```
+wget https://paddle-serving.bj.bcebos.com/models/xpu/vgg19.tar.gz
+tar zxvf vgg19.tar.gz 
 ```
 
 ## RPC Service
@@ -10,7 +20,7 @@ python -m paddle_serving_client.convert --dirname VGG19
 ### Start Service
 
 ```
-python -m paddle_serving_server.serve --model serving_server --port 7702 --use_lite --use_xpu --ir_optim
+python3 -m paddle_serving_server.serve --model serving_server --port 7702 --use_lite --use_xpu --ir_optim
 ```
 
 ### Client Prediction

python/paddle_serving_server/server.py

@@ -386,8 +386,6 @@ class Server(object):
             return
 
         if not os.path.exists(self.server_path):
-            os.system("touch {}/{}.is_download".format(self.module_path,
-                                                       folder_name))
             print('Frist time run, downloading PaddleServing components ...')
 
             r = os.system('wget ' + bin_url + ' --no-check-certificate')
@@ -403,9 +401,10 @@ class Server(object):
                     tar = tarfile.open(tar_name)
                     tar.extractall()
                     tar.close()
+                    open(download_flag, "a").close()
                 except:
-                    if os.path.exists(exe_path):
-                        os.remove(exe_path)
+                    if os.path.exists(self.server_path):
+                        os.remove(self.server_path)
                     raise SystemExit(
                         'Decompressing failed, please check your permission of {} or disk space left.'
                         .format(self.module_path))

tools/scripts/ipipe_py3.sh

@@ -148,7 +148,7 @@ function before_hook() {
     setproxy
     unsetproxy
     cd ${build_path}/python
-    python3.6 -m pip install --upgrade pip
+    python3.6 -m pip install --upgrade pip==20.0.1
     python3.6 -m pip install requests
     python3.6 -m pip install -r requirements.txt -i https://mirror.baidu.com/pypi/simple
     python3.6 -m pip install numpy==1.16.4
@@ -323,7 +323,7 @@ function bert_rpc_cpu() {
     link_data ${data_dir}
     sed -i 's/8860/8861/g' bert_client.py
     python3.6 -m paddle_serving_server.serve --model bert_seq128_model/ --port 8861 > ${dir}server_log.txt 2>&1 &
-    check_result server 3
+    check_result server 5
     cp data-c.txt.1 data-c.txt
     head data-c.txt | python3.6 bert_client.py --model bert_seq128_client/serving_client_conf.prototxt > ${dir}client_log.txt 2>&1
     check_result client "bert_CPU_RPC server test completed"
@@ -338,7 +338,7 @@ function pipeline_imagenet() {
     data_dir=${data}imagenet/
     link_data ${data_dir}
     python3.6 resnet50_web_service.py > ${dir}server_log.txt 2>&1 &
-    check_result server 5
+    check_result server 8
     nvidia-smi
     python3.6 pipeline_rpc_client.py > ${dir}client_log.txt 2>&1
     check_result client "pipeline_imagenet_GPU_RPC server test completed"
@@ -355,7 +355,7 @@ function ResNet50_rpc() {
     link_data ${data_dir}
     sed -i 's/9696/8863/g' resnet50_rpc_client.py
     python3.6 -m paddle_serving_server.serve --model ResNet50_vd_model --port 8863 --gpu_ids 0 > ${dir}server_log.txt 2>&1 &
-    check_result server 5
+    check_result server 8
     nvidia-smi
     python3.6 resnet50_rpc_client.py ResNet50_vd_client_config/serving_client_conf.prototxt > ${dir}client_log.txt 2>&1
     check_result client "ResNet50_GPU_RPC server test completed"
@@ -372,7 +372,7 @@ function ResNet101_rpc() {
     link_data ${data_dir}
     sed -i "22cclient.connect(['127.0.0.1:8864'])" image_rpc_client.py
     python3.6 -m paddle_serving_server.serve --model ResNet101_vd_model --port 8864 --gpu_ids 0 > ${dir}server_log.txt 2>&1 &
-    check_result server 5
+    check_result server 8
     nvidia-smi
     python3.6 image_rpc_client.py ResNet101_vd_client_config/serving_client_conf.prototxt > ${dir}client_log.txt 2>&1
     check_result client "ResNet101_GPU_RPC server test completed"
@@ -482,7 +482,7 @@ function cascade_rcnn_rpc() {
     link_data ${data_dir}
     sed -i "s/9292/8879/g" test_client.py
     python3.6 -m paddle_serving_server.serve --model serving_server --port 8879 --gpu_ids 0 --thread 2 > ${dir}server_log.txt 2>&1 &
-    check_result server 5
+    check_result server 8
     nvidia-smi
     python3.6 test_client.py > ${dir}client_log.txt 2>&1
     nvidia-smi
@@ -499,7 +499,7 @@ function deeplabv3_rpc() {
     link_data ${data_dir}
     sed -i "s/9494/8880/g" deeplabv3_client.py
     python3.6 -m paddle_serving_server.serve --model deeplabv3_server --gpu_ids 0 --port 8880 --thread 2 > ${dir}server_log.txt 2>&1 &
-    check_result server 5
+    check_result server 10
     nvidia-smi
     python3.6 deeplabv3_client.py > ${dir}client_log.txt 2>&1
     nvidia-smi
@@ -516,7 +516,7 @@ function mobilenet_rpc() {
     tar xf mobilenet_v2_imagenet.tar.gz
     sed -i "s/9393/8881/g" mobilenet_tutorial.py
     python3.6 -m paddle_serving_server.serve --model mobilenet_v2_imagenet_model --gpu_ids 0 --port 8881 > ${dir}server_log.txt 2>&1 &
-    check_result server 5
+    check_result server 8
     nvidia-smi
     python3.6 mobilenet_tutorial.py > ${dir}client_log.txt 2>&1
     nvidia-smi
@@ -533,7 +533,7 @@ function unet_rpc() {
     link_data ${data_dir}
     sed -i "s/9494/8882/g" seg_client.py
     python3.6 -m paddle_serving_server.serve --model unet_model --gpu_ids 0 --port 8882 > ${dir}server_log.txt 2>&1 &
-    check_result server 5
+    check_result server 8
     nvidia-smi
     python3.6 seg_client.py > ${dir}client_log.txt 2>&1
     nvidia-smi
@@ -599,7 +599,7 @@ function criteo_ctr_rpc_gpu() {
     link_data ${data_dir}
     sed -i "s/8885/8886/g" test_client.py
     python3.6 -m paddle_serving_server.serve --model ctr_serving_model/ --port 8886 --gpu_ids 0 > ${dir}server_log.txt 2>&1 &
-    check_result server 5
+    check_result server 8
     nvidia-smi
     python3.6 test_client.py ctr_client_conf/serving_client_conf.prototxt raw_data/part-0 > ${dir}client_log.txt 2>&1
     nvidia-smi
@@ -617,7 +617,7 @@ function yolov4_rpc_gpu() {
     sed -i "s/9393/8887/g" test_client.py
     python3.6 -m paddle_serving_server.serve --model yolov4_model --port 8887 --gpu_ids 0 > ${dir}server_log.txt 2>&1 &
     nvidia-smi
-    check_result server 5
+    check_result server 8
     python3.6 test_client.py 000000570688.jpg > ${dir}client_log.txt 2>&1
     nvidia-smi
     check_result client "yolov4_GPU_RPC server test completed"
@@ -634,7 +634,7 @@ function senta_rpc_cpu() {
     sed -i "s/9393/8887/g" test_client.py
     python3.6 -m paddle_serving_server.serve --model yolov4_model --port 8887 --gpu_ids 0 > ${dir}server_log.txt 2>&1 &
     nvidia-smi
-    check_result server 5
+    check_result server 8
     python3.6 test_client.py 000000570688.jpg > ${dir}client_log.txt 2>&1
     nvidia-smi
     check_result client "senta_GPU_RPC server test completed"
@@ -724,7 +724,7 @@ function bert_http() {
     cp vocab.txt.1 vocab.txt
     export CUDA_VISIBLE_DEVICES=0
     python3.6 bert_web_service.py bert_seq128_model/ 8878 > ${dir}server_log.txt 2>&1 &
-    check_result server 5
+    check_result server 8
     curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"words": "hello"}], "fetch":["pooled_output"]}' http://127.0.0.1:8878/bert/prediction > ${dir}client_log.txt 2>&1
     check_result client "bert_GPU_HTTP server test completed"
     kill_server_process
@@ -762,7 +762,7 @@ function grpc_yolov4() {
     link_data ${data_dir}
     echo -e "${GREEN_COLOR}grpc_impl_example_yolov4_GPU_gRPC server started${RES}"
     python3.6 -m paddle_serving_server.serve --model yolov4_model --port 9393 --gpu_ids 0 --use_multilang > ${dir}server_log.txt 2>&1 &
-    check_result server 5
+    check_result server 10
     echo -e "${GREEN_COLOR}grpc_impl_example_yolov4_GPU_gRPC client started${RES}"
     python3.6 test_client.py 000000570688.jpg > ${dir}client_log.txt 2>&1
     check_result client "grpc_yolov4_GPU_GRPC server test completed"