/azure_ml_docker_dev_github

Repo for Azure Machine Learning Service docker development image autobuild

Primary LanguageDockerfile

Azure Machine Learning Service Docker Image for Development

Base Image

This is the docker image for Azure Machine Learning service development task, its baseImage is:

FROM tftwdockerhub/cuda_cudnn_ana_base:latest

Additional installed packages

For Azure Machine Learning service and Data Prep development:

  • Azure ML core SDK: azureml-sdk
  • Azure ML data prep: azureml-dataprep
  • Azure ML extras: azureml-sdk[notebook, etc..]

According to the document, only notebooks, explain, automl, contrib are installed, because databricks and automl_databricks cannot be combined with other components.

  • Azure ML monitoring: azureml-monitoring
  • Azure CLI 2.0 for Azure credential

It is recommended to use lsb_release -cs in the base docker images to check the version of the OS, in this case, it is xenial, so author set this link manually.

For more information, please check:

Docker Registry Repo

  • tftwdockerhub/azure_ml_docker_dev:latest

Usage

on dsvm-gpu virtual machines

sudo docker pull tftwdockerhub/azure_ml_docker_dev:latest

remember the target port is 8889

sudo nvidia-docker run -it -p 8889:8888 -v \<project-dir-path\>:/app tftwdockerhub/azure_ml_docker_dev:latest

In local browser, remember the target port is 8889 and the token string on CLI screen

http://\<vm-ipaddress-or-dns\>:8889/?token=21c5e12xxxxxx