Advanced DL Workbench Configurations

This document contains information on how to run DL Workbench in a more secure mode, the complete list of command-line interface arguments for Python Starter, and additional docker run command-line options.

Python Starter Command-Line Interface

Run this command to get a list of available starting arguments:

openvino-workbench --help

Python Starter arguments:

Argument

Explanation

Default Value

--help

Displays help message and exits the script.

N/A

--image

Specifies the DL Workbench Docker image name.

--force-pull

Force-pull the specified image. Can be used to download a newer version of the DL Workbench image.

false

--ip

Specifies the outside IP to set up the DL Workbench.

0.0.0.0

--port

Maps the Docker container port 5665 to the provided host port to get access to the DL Workbench from a web browser.

5665

--container-name

Specifies the container name to use.

workbench

--detached

Boolean. Enables the detached mode of the Docker container. Container logs will not be visible in the terminal.

false

--enable-gpu

Boolean. Adds a host device to the container. Enables the container to use GPU devices in the DL Workbench.

false

--enable-myriad

Boolean. Mounts directory /dev/bus/usb to the Docker container and adds the rule with allowed devices list to the cgroup. Enables the container to use Intel® Neural Compute Stick 2 devices in the DL Workbench. Cannot be used when running with Vision Accelerator Design with Intel® Movidius™ VPUs.

false

--enable-hddl

Boolean. Adds a host device to the container and mounts directory /var/tmp to the Docker container. Enables the container to use Vision Accelerator Design with Intel® Movidius™ VPUs in the DL Workbench. Cannot be used when running with Intel® Neural Compute Stick 2.

false

--assets-directory

Mounts a provided local folder to the /home/workbench/.workbench directory in the Docker container.

N/A

--http-proxy

Specifies the HTTP proxy in the format http://:@:.

N/A

--https-proxy

Specifies the HTTPS proxy in the format https://:@:.

N/A

--no-proxy

Specifies the URLs that should be excluded from proxying in the format url1,url2,url3.

N/A

--no-jupyter

Disables the Jupyter server.

false

--enable-authentication

Boolean. Turns on authentication for the DL Workbench.

false

--custom-token

Specifies the custom login token for enabled DL Workbench authentication.

N/A

--disable-token-saving

Boolean. Turns off saving of the login token to the file. By default, saving to the file is enabled.

false

--ssl-certificate-name

Specifies the path to the DL Workbench web app TLS certificate in the DL Workbench configuration directory. The file should be placed in “assets_dir” folder. Example:

N/A

--ssl-key-name

Specifies the path to the SSL_CERT certificate private key in the DL Workbench configuration directory.

N/A

--verify-ssl

Indicates whether the SSL_CERT TLS certificate is trusted (on), or either self-signed or untrusted (off).

on

--cloud-service-address

Specifies the URL to a standalone cloud service that provides Intel(R) hardware capabilities for experiments.

N/A

--cloud-service-session-ttl

Specifies the cloud service session time to live.

N/A

--network-name

Specifies the name of a Docker network to run the Docker container in.

--network-alias

Specifies the alias of the DL Workbench container in the network.

N/A

--base-prefix

Specifies the base prefix of the DL Workbench web application.

‘/’

Warning

Before using the --assets-directory argument:

  1. Make sure the directory you pass as a value of --assets-directory has read, write, and execute permissions set for all users. See Troubleshooting for details.

  2. Linux only. Create a group called workbench and add the current user <USERNAME> to it. Use the commands below:

    1. sudo groupadd -g 5665 workbench

    2. sudo usermod -a -G 5665 <USERNAME>

Note

--enable-myriad and --enable-hddl cannot be set simultaneously because Intel® Neural Compute Stick 2 and Intel® Vision Accelerator Design with Intel® Movidius™ VPUs are incompatible and cannot be used in the DL Workbench together.

Docker Run Command-Line Interface

You can run Docker containers with the following arguments:

Argument

Explanation

-p 127.0.0.1:5665:5665

Maps Docker container port 5665 to host port 5665 to get access to the DL Workbench from a web browser.

--device /dev/dri

Add a host device to the container. Enables the container to use GPU devices in the DL Workbench. Linux only.

-v /dev/bus/usb:/dev/bus/usb

Mounts directory /dev/bus/usb to the Docker container. Enables the container to use Intel® Neural Compute Stick 2 devices in the DL Workbench. Linux only. Cannot be used when running with Vision Accelerator Design with Intel® Movidius™ VPUs.

--device-cgroup-rule='c 189:\* rmw'

Adds the rule with allowed devices list to the cgroup. Enables the container to use Intel® Neural Compute Stick 2 devices in the DL Workbench. Linux only. Cannot be used when running with Vision Accelerator Design with Intel® Movidius™ VPUs.

--device /dev/ion:/dev/ion

Add a host device to the container. Enables the container to use Vision Accelerator Design with Intel® Movidius™ VPUs in the DL Workbench. Linux only.*Cannot be used when running with Intel® Neural Compute Stick 2.*

-v /var/tmp:/var/tmp

Mounts directory /var/tmp to the Docker container. Enables the container to use Vision Accelerator Design with Intel® Movidius™ VPUs devices in the DL Workbench. Linux only. Cannot be used when running with Intel® Neural Compute Stick 2.

-it

Enables the interactive mode of the Docker container. Set to the Docker image name workbench . Docker container allows interactive processes in the DL Workbench terminal. To stop a container in this mode, press Ctrl + C .

-d

Enables the detached mode of the Docker container. Set to the Docker image name workbench . Docker container runs in the background of your terminal and does not receive input or display output. To stop a container in this mode, run the docker stop workbench command.

--volume ~/.workbench:/home/workbench/.workbench

Mounts a local folder named ~/.workbench to the /home/workbench/.workbench directory in the Docker* container

-e https_proxy=<https-proxy> -e http_proxy=<http-proxy> -e no_proxy=<no-proxy>

Optional. If you are behind a corporate proxy, set environment variables.

-e ENABLE_AUTH=1

Boolean. Turns on authentication for the DL Workbench.

-e CUSTOM_TOKEN=<custom-token>

Specifies a custom login token for enabled DL Workbench authentication. By default, a login token is generated automatically.

-e SAVE_TOKEN_TO_FILE=0

Boolean. Turns off saving of the login token to the file. By default, saving to the file is enabled.

-e BASE_PREFIX=<base-prefix>

Specifies the base prefix of the DL Workbench web application. Default value is ‘/’.

For other options, like launching the DL Workbench container or restarting the container, see Docker Container section.

Run DL Workbench Securely

By default, the DL Workbench is started without authentication settings. If you want to protect your sensitive data, like the model details or performance information, you can enable authentication in DL Workbench and use a self-signed certificate.

Disable JupyterLab inside DL Workbench

Jupyter notebooks running in the same Docker container as the DL Workbench can impact inference results of experiments inside the DL Workbench. To get more accurate performance information, you might want to disable JupyterLab Environment. Use the DISABLE_JUPYTER or --no-jupyter argument when starting the DL Workbench. For example:

docker run -p 127.0.0.1:5665:5665 --name workbench -e DISABLE_JUPYTER=1 -it openvino/workbench:latest
openvino-workbench --image openvino/workbench:2022.1 --no-jupyter