15
Intel® Deep Learning SDK Training Tool Installation Guide Copyright © 2016 Intel Corporation All Rights Reserved Revision: 002

Intel(R) Deep Learning SDK Training Tool Installation Guide

  • Upload
    others

  • View
    24

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Intel(R) Deep Learning SDK Training Tool Installation Guide

Intel® Deep Learning SDK Training Tool

Installation Guide

Copyright © 2016 Intel Corporation

All Rights Reserved

Revision: 002

Page 2: Intel(R) Deep Learning SDK Training Tool Installation Guide

Intel® Deep Learning SDK Training Tool

2 Installation Guide

Contents 1 Introduction ............................................................................................................. 3

2 Prerequisites ............................................................................................................ 4

3 Installing from a Microsoft Windows* or Apple macOS* Machine .................................... 5 3.1 License Agreement ......................................................................................................................6 3.2 Client settings ..............................................................................................................................6 3.3 Server Settings .............................................................................................................................7

3.3.1 Linux server settings .................................................................................... 7 3.3.2 Amazon EC2* Server Settings ....................................................................... 9 3.3.3 Settings for local installation on your Apple macOS* machine .......................... 11

3.4 Installation ................................................................................................................................ 13 3.5 Completed ................................................................................................................................ 13

4 Installing Locally on a Linux* Machine ........................................................................ 15

Page 3: Intel(R) Deep Learning SDK Training Tool Installation Guide

Introduction

Installation Guide 3

1 Introduction The Training Tool operates in a client-server architecture, where the web client works in connection with a remote Unix-based server. Therefore the tool supports various installation options:

• Installing using the GUI-based installer on your Microsoft Windows* machine, to install the Training Tool on Linux* or Amazon Cloud* servers

• Installing using the GUI-based installer on your Apple macOS* machine, to install the Training Tool on your Mac* or Linux* or Amazon Cloud* servers

• Running the installation script, to install the Training Tool locally on your Linux machine.

You do not need to install any additional software manually as the installation package consists of a Docker* container that contains all necessary components including the Intel® Distribution for Caffe* with all prerequisites and provides the environment for running the Training Tool.

Page 4: Intel(R) Deep Learning SDK Training Tool Installation Guide

Intel® Deep Learning SDK Training Tool

4 Installation Guide

2 Prerequisites Make sure you comply with the following requirements before beginning the installation of the Intel® Deep Learning SDK Training Tool. The requirements are also available in the Release Notes document that can be found online and in your installation package.

• Hardware: Optimized for Intel® Xeon® and Intel® Core™ processors • Operating Systems:

• Ubuntu* 14.04 or higher (64-bit) • CentOS* 7 (64-bit) • macOS* 10.10.3 or higher (64-bit) • CoreOS* (on an Amazon EC2* server instance)

• Software: Google Chrome* browser version 50 or higher, which is used to access the Training Tool web user interface

• Free Disk Space: 1 GB.

Linux* machine • A Linux Ubuntu* or Cent OS* machine accessible through a SSH connection • Root privileges to run the installation script on the Linux machine

Apple macOS* machine • Docker* for Mac version 1.12 or higher installed • VirtualBox* version lower than 4.3.30 must not be installed because it is incompatible with Docker

for Mac

Amazon EC2 server • Valid Amazon Web Service (AWS) account • Valid Amazon EC2 private secure key

To generate EC2 key pairs for a specific region, use the Key Pairs tab in the AWS EC2 Dashboard.

NOTE: While installing the Training Tool, make sure that the Default region name value matches the region specified for key generation.

To check EC2 server instance state, use the Instances tab.

For information about creating and configuring your AWS account, see Getting Started with AWS.

Page 5: Intel(R) Deep Learning SDK Training Tool Installation Guide

Installing from a Microsoft Windows* or Apple macOS* Machine

Installation Guide 5

3 Installing from a Microsoft Windows* or Apple macOS* Machine To install the Intel® Deep Learning SDK Training Tool from a Microsoft Windows* or Apple macOS* machine, download the installation package from https://software.intel.com/deep-learning-sdk, unpack and launch the TrainingToolInstaller executable file to start the wizard.

The macOS and Windows installation wizards look similar and contain the same steps.

NOTE: Local installation is only available for Apple Mac* machines and can be done using the macOS installer.

The wizard launches the following steps that guide you through the installation process and advance as you click the Next button. The installation process includes the following steps.

Page 6: Intel(R) Deep Learning SDK Training Tool Installation Guide

Intel® Deep Learning SDK Training Tool

6 Installation Guide

3.1 License Agreement Read carefully and accept the License Agreement to continue with the installation.

3.2 Client Settings Specify the following settings for the client part of the Training Tool:

• Training Tool User Name – User name to access the Training Tool web interface • Training Tool password – Password to access the Training Tool web interface.

Page 7: Intel(R) Deep Learning SDK Training Tool Installation Guide

Installing from a Microsoft Windows* or Apple macOS* Machine

Installation Guide 7

3.3 Server Settings Select the server machine to install the Training Tool and configure the installation parameters including network and security settings. Specify all required field values - marked with a red asterix - and modify default ones if needed according to the server type.

3.3.1 Linux* Server Settings To install the Training Tool on your own Linux* server machine, select the Linux Server option and specify the following mandatory settings:

Page 8: Intel(R) Deep Learning SDK Training Tool Installation Guide

Intel® Deep Learning SDK Training Tool

8 Installation Guide

• Server name or IP – The address of the Linux machine on which the Training Tool should be

installed • User name (with root access) – User name with root privileges on the Linux server • User password –The password of the above user account. These credentials are needed for user

authentication in the installation process • Connection security level – Secured or non-secured connection between the client and server.

If you select the non-secured connection, you should be aware that any communication between the client and the server is not encrypted and any tapping to the network can reveal the content transmitted or expose you to man-in-the-middle attacks. User passwords are still secured and protected in this case. You should not use this option in public network environment. To use the secured (HTTPS) connection type, provide the secure certificate using the following option:

• Upload a certificate. Use this option to upload your user owned certificate issued by a certificate authority. Choose and load the certificate file and provide the certificate password if

Page 9: Intel(R) Deep Learning SDK Training Tool Installation Guide

Installing from a Microsoft Windows* or Apple macOS* Machine

Installation Guide 9

required. Make sure the certificate domain name is matching the domain name you gave for the server you are installing.

In addition to the mandatory fields you can configure Advanced settings if needed:

• Private key file for user authentication – The private key used for user authentication when password authentication is not allowed on the Linux server

• If the connection between installing machine and the server goes through a proxy, set the corresponding proxies accordingly: • Proxy server for HTTP – Set the HTTP proxy server address and port number • Proxy server for HTTPS – Set the HTTPS proxy server address and port number

• Mount file system path – Linux file system path which is to be mounted as a volume inside the Docker container

• Web application port - Network port to access the Training Tool web interface. • Web application port for interactive notebook - Network port to access the Jupyter Notebook*

web interface.

Once you complete server setting, you can check the connection to the server by pressing the Test connection button. If the server is accessible, the test will result in the Connection successful status:

3.3.2 Amazon EC2* Server Settings To install the Training tool on an Amazon EC2* instance, select the Amazon Cloud Server option and fill in the following fields:

Page 10: Intel(R) Deep Learning SDK Training Tool Installation Guide

Intel® Deep Learning SDK Training Tool

10 Installation Guide

• Cluster name – Name of the cluster that will be created and used to install the tool • Volume size (GB) – Cloud capacity (in gigabytes) dedicated to EC2 server instance on which the

tool is to be installed • Instance type – Server instance type that includes a combination of CPU, memory, storage, and

networking capacity • Private key file for authentication – Your Amazon EC2 private secure key file to log in to the EC2

instance

• Proxy server for HTTPS – Set the proxy server address and port to provide HTTPS access from your Linux server to AWS

• Proxy server for SSH - Set the proxy server address and port number - if required - to establish an SSH connection from your local machine to the EC2 server instance

• Proxy server type for SSH – Select the proxy server type out of the drop-down list of socks4, socks5, HTTPS, telnet or local

• AWS access key ID – Your AWS access key ID (20 symbols)

Page 11: Intel(R) Deep Learning SDK Training Tool Installation Guide

Installing from a Microsoft Windows* or Apple macOS* Machine

Installation Guide 11

• AWS secret access – Your AWS secret access key (40 symbols) • Default region name – Your AWS region. Make sure that the Default region name value matches

the region you specified for generating your EC2 key pair. • Connection security level – Select between the secured (https) or non-secured connection types

between the client and server. To use the secured (https) connection type, provide the secure certificate using the following option: • Upload a certificate. Use this option to upload your user owned certificate issued by a

certificate authority. Choose and load the certificate file and provide the certificate password if required. Make sure the certificate domain name is matching the domain name you gave for the server you are installing.

3.3.3 Settings for Local Installation on Your Apple macOS* Machine

NOTE: This option is available only in the Apple macOS* version of the installer.

To install the Training Tool locally on your Apple macOS machine, specify the following server settings:

Page 12: Intel(R) Deep Learning SDK Training Tool Installation Guide

Intel® Deep Learning SDK Training Tool

12 Installation Guide

• Connection security level – Select between the secured (https) or non-secured connection types

between the client and server. To use the secured (https) connection type, provide the secure certificate using the following option: • Upload a certificate. Use this option to upload your user owned certificate issued by a

certificate authority. Choose and load the certificate file and provide the certificate password if required. Make sure the certificate domain name is matching the domain name you gave for the server you are installing.

• Proxy server for HTTP – Set the HTTP proxy server address and port number for Internet access inside the Docker container

• Proxy server for HTTPS – Set the HTTPS proxy server address and port number for Internet access inside the Docker container

• Mount file system path – macOS file system path which is to be mounted as a volume inside the Docker* container

• Web application port - Network port to access the Training Tool web interface. • Web application port for interactive notebook - Network port to access the Jupyter Notebook*

web interface.

Page 13: Intel(R) Deep Learning SDK Training Tool Installation Guide

Installing from a Microsoft Windows* or Apple macOS* Machine

Installation Guide 13

3.4 Installation Click the Install button to begin the installation. The Installing panel comes up to show the progress of the installation with a progress bar that depicts the current status and the overall completeness:

When the installation is complete (when the completed percentage goes all the way to100%), the Next button will be activated; click it to complete the installation.

3.5 Installation Complete Congratulations! You have installed the Intel® Deep Learning SDK Training Tool.

Page 14: Intel(R) Deep Learning SDK Training Tool Installation Guide

Intel® Deep Learning SDK Training Tool

14 Installation Guide

Click the Open now button to open the Training Tool web interface in your default browser, or copy the URL into Google Chrome* browser address field. By clicking the Download link you can download the latest version of the Google Chrome browser. Hit the Close button to close the window.

Page 15: Intel(R) Deep Learning SDK Training Tool Installation Guide

Installing Locally on a Linux* Machine

Installation Guide 15

4 Installing Locally on a Linux* Machine You can install the Intel® Deep Learning SDK Training Tool on a Linux* operating system using the install_training_tool.sh installation script. The script is available at https://software.intel.com/deep-learning-sdk. Download, run and follow the installation instructions:

./install_training_tool.sh

Optionally, you can specify the following available options:

-volume <path> - Linux file system path which will be mounted as a volume inside the Docker* container

-toolpassword <password> - admin password to access the Training Tool web interface

-toolport <port> - network port to access the Training Tool web interface

-pfxcertificate <true/false> - pass true if you want to load a .pfx file as a certificate

-pfxcertificatepath <path> - path to the certificate .pfx file

-pfxcertificatepassword <password> - password to access the .pfx file

-httpproxy <proxy> - proxy server for HTTP

-httpsproxy <proxy> - proxy server for HTTPS

-help - print help message

NOTE: While the toolpassword parameter is mandatory and must be set to continue the installation, other parameters are all auxiliary and can be omitted.

For example, to install the Training Tool using “123” as a password and port number 8081, type the following command:

./install_training_tool.sh –toolpassword 123 -toolport 8081