Merge branch 'develop' into toolchain0.7

README.md

@@ -60,6 +60,7 @@ This chapter guides you through the installation and deployment steps. It is str
 - [Deploy Paddle Serving on Kubernetes](doc/Run_On_Kubernetes_CN.md)
 - [Deploy Paddle Serving with Security gateway(Chinese)](doc/Serving_Auth_Docker_CN.md)
 - [Deploy Paddle Serving on more hardwares](doc/Run_On_XPU_EN.md)
+- [Docker Images](doc/Docker_Images_EN.md)
 - [Latest Wheel packages](doc/Latest_Packages_CN.md)
 
 > Use

README_CN.md

@@ -56,6 +56,7 @@ Paddle Serving依托深度学习框架PaddlePaddle旨在帮助深度学习开发
 - [在Kuberntes集群上部署Paddle Serving](doc/Run_On_Kubernetes_CN.md)
 - [部署Paddle Serving安全网关](doc/Serving_Auth_Docker_CN.md)
 - [在异构硬件部署Paddle Serving](doc/Run_On_XPU_CN.md)
+- [Docker镜像](doc/Docker_Images_CN.md)
 - [最新Wheel开发包](doc/Latest_Packages_CN.md)
 
 > 使用

core/cube/cube-builder/tool/kv_to_seqfile.py

@@ -54,13 +54,21 @@ def kv_to_seqfile():
         finally:
             fp.close()
         for line in lines:
-            line_list = line.split(':')
+            line_list = line.split()
+            if len(line_list) < 1:
+                continue
             key = int(line_list[0])
-            value = str(line_list[1]).replace('\n', '')
+            show = int(line_list[1])
+            click = int(line_list[2])
+            values = [float(x) for x in line_list[3:]]
+
+           # str(line_list[1]).replace('\n', '')
             res.append(dict)
             key_bytes = struct.pack('Q', key)
-            row_bytes = struct.pack('%ss' % len(value), value)
-            print key, ':', value, '->', key_bytes, ':', row_bytes
+            row_bytes = ""
+            for v in values:
+                row_bytes += struct.pack('f', v)
+            print key, ':', values, '->', key_bytes, ':', row_bytes
             writer.write(key_bytes, row_bytes)
     f.close()
     write_donefile()

core/predictor/tools/seq_reader.cpp

@@ -55,7 +55,14 @@ void printSeq(std::string file, int limit) {
 
     total_count++;
     int64_t value_length = record.record_len - record.key_len;
-    std::cout << "key: " << key << " , value: " << string_to_hex(record.value.c_str()) << std::endl; 
+    float *data_ptr = new float[record.value.size() / 4];
+    memcpy(data_ptr, record.value.data(), record.value.size()); 
+    std::cout << "key: " << key << " , value: " << string_to_hex(record.value.c_str()) << std::endl;
+    for (int i =0; i < record.value.size() / 4; ++i) {
+      std::cout << data_ptr[i] << " ";
+    } 
+    std::cout << std::endl;
+    delete(data_ptr);
     if (total_count >= limit) {
         break;
     }

doc/Run_On_XPU_CN.md

@@ -3,6 +3,9 @@
 
 Paddle Serving支持使用百度昆仑芯片进行预测部署。目前支持在百度昆仑芯片和arm服务器（如飞腾 FT-2000+/64）, 或者百度昆仑芯片和Intel CPU服务器，上进行部署，后续完善对其他异构硬件服务器部署能力。
 
+## 安装Docker镜像
+我们推荐使用docker部署Serving服务，在xpu环境下可参考[Docker镜像](Docker_Images_CN.md)文档安装xpu镜像，并进一步完成编译、安装和部署等任务。
+
 ## 编译、安装
 基本环境配置可参考[该文档](Compile_CN.md)进行配置。下面以飞腾FT-2000+/64机器为例进行介绍。
 ### 编译

doc/Run_On_XPU_EN.md

@@ -5,6 +5,9 @@
 Paddle serving supports deployment using Baidu Kunlun chips. Currently, it supports deployment on the ARM CPU server with Baidu Kunlun chips
  (such as Phytium FT-2000+/64), or Intel CPU with Baidu Kunlun chips. We will improve
  the deployment capability on various heterogeneous hardware servers in the future. 
+ 
+## Install docker images
+We recommend using the docker deployment service. In the xpu environment, you can refer to the [Docker image document](Docker_Images_EN.md) to install the xpu image, and further complete tasks such as construction, installation, and deployment.
 
 ## Compilation and installation
 Refer to [compile](./Compile_EN.md) document to setup the compilation environment. The following is based on FeiTeng FT-2000 +/64 platform.