This document will explain how to deploy an Azure DSVM and set up the environment for Active Learning.

Create an SSH Key on your local machine. The following will create a key in your ~/.ssh/act-learn-key location.

$ ssh-keygen -f ~/.ssh/act-learn-key -t rsa -b 2048

Secondly edit the environment variables in the dsvm_config.sh script with your own values. For instance:

RESOURCE_GROUP=MyAzureResourceGroup
# VM config
VM_SKU=Standard_NC6 #Make sure VM SKU is available in your resource group's region 
VM_IMAGE=microsoft-ads:linux-data-science-vm-ubuntu:linuxdsvmubuntu:latest
VM_DNS_NAME=mytestdns
VM_NAME=myvmname
VM_ADMIN_USER=johndoe
VM_SSH_KEY=~/.ssh/act-learn-key.pub

Lastly execute the deploy_dsvm.sh with your edited config file as a parameter. Note that the Azure CLI is required. Install here if needed.

$ sh deploy_dsvm.sh config/dsvm_config.sh

We provide a module that will copy over a shell script to your DSVM and execute the shell script to setup an active learning environment.

We require that your SSH key be added to the SSH agent. To add your SSH key to the SSH agent use the ssh-add command

$ ssh-add -K ~/.ssh/act-learn-key

To copy and execute the shell script on the DSVM use the following command

$ python setup-tensorflow.py --host admin@127.0.0.1 -k ~/.ssh/act-learn-key -s setup-tensorflow.sh

Note that in the host argument admin@127.0.0.1 section is the DSVM Admin name and admin@127.0.0.1 is the IP address of the DSVM.