diff --git a/mindspore/lite/test/run_benchmark_nets.sh b/mindspore/lite/test/run_benchmark_nets.sh
index cc0bca64e1..11952cece1 100644
--- a/mindspore/lite/test/run_benchmark_nets.sh
+++ b/mindspore/lite/test/run_benchmark_nets.sh
@@ -1,77 +1,76 @@
 #!/bin/bash
 
-#获取相关输入参数
-#举例：sh run_benchmark_nets.sh -a /home/temp_test -c /home/temp_test -m /home/temp_test/models -d "8KE5T19620002408"
+#example：sh run_benchmark_nets.sh -a /home/temp_test -c /home/temp_test -m /home/temp_test/models -d "8KE5T19620002408"
 while getopts "a:c:m:d:" opt
 do
     case $opt in
         a)
 		arm_path=$OPTARG
-        echo "参数arm_path的值$OPTARG"
+        echo "arm_path is $OPTARG"
         ;;
         c)
 		convertor_path=$OPTARG
-        echo "参数convertor_path的值$OPTARG"
+        echo "convertor_path is $OPTARG"
         ;;
         m)
 		models_path=$OPTARG
-        echo "参数models_path的值$OPTARG"
+        echo "models_path is $OPTARG"
         ;;		
         d)
 		device_id=$OPTARG
-        echo "参数device_id的值$OPTARG"
+        echo "device_id is $OPTARG"
         ;;
         ?)
-        echo "未知参数"
+        echo "unknown para"
         exit 1;;
     esac
 done
 
-#将编译好的arm包先放在如下目录进行调试 
+#unzip arm 
 cd $arm_path
 tar -zxf MSLite-*-linux_arm64.tar.gz
 
 
-#部署模型转换工具 
+#unzip convertor 
 cd $convertor_path
 tar -zxf MSLite-*-linux_x86_64.tar.gz
 cd $convertor_path/MSLite-*-linux_x86_64
 cp converter/converter_lite ./
 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:./lib/:./third_party/protobuf/lib
 
-#获取原始模型提前放置在$models_path/
+#the original model's path: $models_path/
 
 
-#进行模型转换
+#convert the models
 cd $convertor_path/MSLite-*-linux_x86_64
-#模型1：
+#model1：
 ./converter_lite  --fmk=CAFFE --modelFile=$models_path/test.prototxt --outputFile=$models_path/test --weightFile=$models_path/test.caffemodel 
-#模型2：
+#model2：
 ./converter_lite  --fmk=TFLITE --modelFile=$models_path/hiai_bigmodel_ghost_2_1_no_normalized_no_trans_tflite.tflite --outputFile=$models_path/hiai_bigmodel_ghost_2_1_no_normalized_no_trans_tflite 
 
 
-#推送到手机上执行benchmark：
+#push to the arm and run benchmark：
 
-#一：复制到手机所在的机器上 
+#first：copy to the server which connected to the phone
 mkdir -p ./benchmark_test
 cp  $arm_path/MSLite-0.6.0-linux_arm64/lib/libmindspore-lite.so ./benchmark_test/libmindspore-lite.so
 cp  $arm_path/MSLite-0.6.0-linux_arm64/benchmark/benchmark ./benchmark_test/benchmark
 
-#复制模型到连接手机服务器所在目录：
-#模型1：
+#copy the models：
+#model1：
 cp  $models_path/test.ms ./benchmark_test/
-#模型2：
+#model2：
 cp  $models_path/hiai_bigmodel_ghost_2_1_no_normalized_no_trans_tflite.ms ./benchmark_test/
 
-#二：adb 推送到手机上
+#second：adb push to the phone
 adb -s $device_id push ./benchmark_test /data/local/tmp/
 
-#三：执行adb命令，运行推理，获取返回值判断结果；
+#third：run adb ,run session ,check the result:
 echo 'cd  /data/local/tmp/' > adb_cmd.txt
 echo 'chmod 777 benchmark' >> adb_cmd.txt
-#模型1：
+#model1：
 echo 'export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/data/local/tmp/;./benchmark --modelPath=test.ms' >> adb_cmd.txt
-#模型2：
+#model2：
 echo 'export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/data/local/tmp/benchmark_test;./benchmark --modelPath=hiai_bigmodel_ghost_2_1_no_normalized_no_trans_tflite.ms' >> adb_cmd.txt
 
 adb -s $device_id shell < adb_cmd.txt