แก้ไข

แชร์ผ่าน

Azure Data Science Virtual Machine release notes

  • บทความ

This article describes Azure Data Science Virtual Machine releases. For a full list of included tools, along with version numbers, visit this resource.

Because of rapidly evolving needs and packages updates, we target new releases of Azure Data Science Virtual Machine for Windows and Ubuntu images every month.

Azure portal users can find the latest image available for provisioning the Data Science Virtual Machine. For CLI or Azure Resource Manager (ARM) users, we keep images of individual versions available for 12 months. After that period, specific image versions are no longer available for provisioning.

Visit the list of known issues to learn about known bugs and workarounds.

October 21, 2024

Data Science Virtual Machine – Ubuntu 22.04

Description:

DSVM-Ubuntu-22.04 image is a tailored Data Science Virtual Machine built on the robust Ubuntu 22.04 platform. It comes pre-installed with a comprehensive suite of popular data science and AI tools, including deep learning frameworks, Jupyter notebooks, and visualization libraries. Optimized for seamless development and deployment, this image offers enhanced compatibility with modern hardware, enabling data scientists, machine learning engineers, and AI researchers to accelerate their workflows efficiently. Whether you're building models, running simulations, or analyzing large datasets, DSVM-Ubuntu-22.04 empowers you with the tools and performance needed to drive innovation.

  • OS: Ubuntu-22.04
  • ImageVersion: 24.10.02
  • SDK-version: 1.57.0
  • Python Version: 3.10.8

DSVM 22.04 has all the latest packages that we had installed and shipped with 20.04

October 18, 2024

Data Science Virtual Machine – Ubuntu 20.04

Version 24.10.01

  • SDK 1.57.0
  • NVIDIA 535.183.01
  • Cuda cuda_12.2.r12
  • Python Version 3.10.8

Data Science Virtual Machine - Windows 2022

Version 24.09.26

  • SDK 1.57.0

Data Science Virtual Machine - Windows 2019

Version 24.10.07

  • SDK 1.57.0

June 17, 2024

Data Science Virtual Machine - Windows 2022

Version 24.05.20

  • SDK 1.56.0

Data Science Virtual Machine - Windows 2019

Version 24.05.20

  • SDK 1.56.0

Data Science VM – Ubuntu 20.04

Version 24.05.24

  • SDK 1.56.0

May 4, 2024

Data Science Virtual Machine - Windows 2022

Version 24.05.03

  • SDK 1.55
  • azureml-dataset-runtime version 1.55.0
  • azureml-train-automl-client
  • azureml-dataset-runtime [fuse,pandas] version 1.55.0
  • GitPython version 3.1.41
  • pyarrow version 14.0.2

Resolved issue: Office365ProPlus installation

December 20, 2023

Data Science VM – Ubuntu 20.04

Version 23.12.18

Primary changes:

  • numpy version 1.22.3
  • pytz version 2022.6
  • torch version 1.12.0
  • certifi version 2023.7.2
  • azure-mgmt-network to version 25.1.0
  • scikit-learn version 1.0.2
  • scipy version 1.9.2
  • accuracy
  • pickle5
  • pillow version 10.1.0
  • experimental
  • ipykernel version 6.14.0
  • en_core_web_sm

December 18, 2023

Data Science Virtual Machine - Windows 2019

Version 23.12.11

Primary changes:

  • SDK 1.54.0
  • numba
  • Scipy
  • azure-core to version 1.29.4
  • azure-identity to version 1.14.0
  • azure-storage-queue to version 12.7.2

December 5, 2023

The Data Science Virtual Machine (DSVM) offering for Data Science VM – Windows 2022 is now generally available in the marketplace.

Version 23.11.23

Primary changes:

  • SDK 1.54.0
  • numba
  • Scipy
  • azure-core to version 1.29.4
  • azure-identity to version 1.14.0
  • azure-storage-queue to version 12.7.2

July 26, 2023

A new Data Science Virtual Machine (DSVM) offering for Data Science VM – Windows 2022 (Preview) is now live in the marketplace.

Version 23.06.25

Primary changes:

  • SDK 1.51.0

April 26, 2023

Data Science VM – Ubuntu 20.04

Version 23.04.24

Primary changes:

  • SDK 1.50.0
  • Dotnet upgraded to 6.0 SDK
  • PyTorch GPU functionality fixed in `azureml_py38_PT_and_TF environment.
  • Blobfuse upgraded to blobfuse2

April 4, 2023

Data Science Virtual Machine - Windows 2019

Version: 23.03.31

Primary changes:

  • SDK 1.49
  • Cuda drivers upgraded to 11.4
  • PyTorch GPU functionality fixed on azureml_py38 and azureml_py28_PT_and_TF environments
  • Dotnet upgraded to 6.0

January 10, 2023

Data Science VM – Ubuntu 20.04

Version: 23.01.06

Primary changes:

  • Added R package "ranger"
  • Pinned pandas==1.1.5 and numpy==1.23.0 in azureml_py38 environment

November 30, 2022

Data Science VM – Ubuntu 20.04

Version: 22.11.25

Primary changes:

  • Azure Machine Learning SDK V2 samples included
  • Ray to version 2.0.0
  • Added clock, recipes R packages
  • azureml-core to version 1.47.0
  • azure-ai-ml to version 1.1.1

Data Science Virtual Machine - Windows 2019

Version: 22.11.27

Primary changes:

  • Azure Machine Learning SDK V2 samples included
  • RScirpt environment path alignment
  • Ray version 2.0.0 package added to azureml_py38 and azureml_py38_PT_TF environments.
  • azureml-core to version 1.47.0
  • azure-ai-ml to version 1.1.1

September 20, 2022

Announcement: Beginning October 1, 2022, the Ubuntu 18 Data Science Virtual Machine (DSVM) will not be available on the marketplace. We recommend that users switch to the Ubuntu 20 DSVM as we continue to ship updates/patches on our latest Data Science VM – Ubuntu 20.04

Users who use the Azure Resource Manager (ARM) template/virtual machine scale, set to deploy the Ubuntu DSVM machines, should set the configuration to

Offer SKU
ubuntu-2004 2004 for Gen1 or 2004-gen2 for Gen2 VM sizes

instead of:

Offer SKU
ubuntu-1804 1804 for Gen1 or 1804-gen2 for Gen2 VM sizes

Note

There's no problem for existing customers who still use the Ubuntu-18 DSVM, as of our October 2022 update. However, the deprecation plan is scheduled for December 2022. We recommend that you switch to Ubuntu-20 DSVM at your earliest convenience.

September 19, 2022

Data Science VM – Ubuntu 20.04

Version 22.09.19

Primary changes:

  • .NET Framework to version 3.1.423
  • Azure Cli to version 2.40.0
  • Intelijidea to version 2022.2.2
  • Microsoft Edge Browser to version 107.0.1379.1
  • Nodejs to version v16.17.0
  • Pycharm to version 2022.2.1

Environment Specific Updates:

azureml_py38:

  • azureml-core to version 1.45.0

py38_default:

  • Jupyter Lab to version 3.4.7
  • azure-core to version 1.25.1
  • keras to version 2.10.0
  • tensorflow-gpu to version 2.10.0

September 12, 2022

Data Science Virtual Machine - Windows 2019

Version 22.09.06

Primary changes:

  • Base OS level image updates.

August 16, 2022

Data Science VM – Ubuntu 20.04

Version 22.08.11

Primary changes:

  • Jupyterlab upgraded to version 3.4.5
  • matplotlib, azureml-mlflow added to sdkv2 environment.
  • Jupyterhub spawner reconfigured to root environment.

July 28, 2022

Data Science VM – Ubuntu 20.04

Version 22.07.19

Primary changes:

  • Updated Azure Cli to version 2.38.0

  • Updated Nodejs to version v16.16.0

  • Updated Scala to version 2.12.15

  • Updated Spark to version 3.2.2

  • MMLSpark notebook features v0.10.0

  • 4 other R libraries:

  • Added new Azure Machine Learning Environment azureml_310_sdkv2

Data Science Virtual Machine - Windows 2019

Version 22.07.18

Primary changes:

  • General OS level updates.

July 11, 2022

Data Science VM – Ubuntu 18.04 Data Science VM – Ubuntu 20.04

Version 22.07.08

Primary changes:

  • Minor bug fixes.

June 28, 2022

Data Science Virtual Machine - Windows 2019 Data Science VM – Ubuntu 20.04

Version 22.06.10

Data Science VM – Ubuntu 18.04

Version 22.06.13

Primary changes:

  • Remove Rstudio software tool from Data Science Virtual Machine (DSVM) images.

May 17, 2022

Data Science VM – Ubuntu 20.04

Version 22.05.11

Primary changes:

  • Upgraded log4j(v2) to version 2.17.2

April 29, 2022

Data Science VM – Ubuntu 18.04 Data Science VM – Ubuntu 20.04

Version 22.04.27

Primary changes:

  • Plotly and summarytools R studio extensions runtime import fix.
  • Cudatoolkit and CUDNN upgraded to 13.1 and 2.8.1 respectively.
  • Fix Python 3.8 - Azure Machine Learning notebook run, pinned matplotlib to 3.2.1 and cycler to 0.11.0 packages in Azureml_py38 environment.

April 26, 2022

Data Science Virtual Machine - Windows 2019

Version: 22.04.21

Primary changes:

  • Plotly R studio extension patch.
  • Update Rscript env path to support latest R studio version 4.1.3.

April 14, 2022

A new Data Science Virtual Machine (DSVM) offering for Data Science VM – Ubuntu 20.04 is currently live in the marketplace.

Version: 22.04.05

April 04, 2022

A new image for Data Science VM – Ubuntu 18.04

Version: 22.04.01

Primary changes:

  • Updated R environment - added these libraries:

    • Cluster
    • Devtools Factoextra
    • GlueHere
    • Ottr
    • Paletteer
    • Patchwork
    • Plotly
    • Rmd2jupyter
    • Scales
    • Statip
    • Summarytools
    • Tidyverse
    • Tidymodels
    • Testthat

  • Further Log4j vulnerability mitigation - although not used, we moved all log4j to version v2, we removed old log4j jars1.0, and we moved log4j version 2.0 jars

  • Azure CLI to version 2.33.1

  • Fixed jupyterhub access issue using public ip address

  • Redesign of Conda environments - as we proceed with alignment and refinement of the Conda environments, we created:

    • azureml_py38: environment based on Python 3.8 with preinstalled Azure Machine Learning SDK that also contains the AutoML environment
    • azureml_py38_PT_TF: an additional azureml_py38 environment, preinstalled with latest TensorFlow and PyTorch
    • py38_default: default system environment based on Python 3.8
    • We removed the
      • azureml_py36_tensorflow
      • azureml_py36_pytorch
      • py38_tensorflow
      • py38_pytorch environments

March 18, 2022

Data Science Virtual Machine - Windows 2019

Version: 22.03.09

Primary changes:

  • Updated R environment - added these libraries:

    • Cluster
    • Devtools Factoextra
    • GlueHere
    • Ottr
    • Paletteer
    • Patchwork
    • Plotly
    • Rmd2jupyter
    • Scales
    • Statip
    • Summarytools
    • Tidyverse
    • Tidymodels
    • Testthat

  • Further Log4j vulnerability mitigation - although not used, we moved all log4j to version v2, we removed old log4j jars1.0, and we moved log4j version 2.0 jars.

  • Azure CLI to version 2.33.1

  • Redesign of Conda environments - as proceed with alignment and refinement of the Conda environments, we created:

    • azureml_py38: environment based on Python 3.8 with preinstalled Azure Machine Learning SDK containing also AutoML environment
    • azureml_py38_PT_TF: complementary environment azureml_py38 with preinstalled with latest TensorFlow and PyTorch
    • py38_default: default system environment based on Python 3.8
    • we removed azureml_py36_tensorflow, azureml_py36_pytorch, py38_tensorflow and py38_pytorch environments.

March 9, 2022

Data Science Virtual Machine - Windows 2019

Version: 21.12.03

The Windows 2019 Data Science Virtual Machine (DSVM) is supported under publisher: microsoft-dsvm, offer ID: dsvm-win-2019, plan ID/SKU ID: winserver-2019

Users who use the Azure Resource Manager (ARM) template / virtual machine scale, set to deploy the Windows DSVM machines, should configure the SKU with winserver-2019 instead of server-2019, since we continue to ship updates to Windows DSVM images on the new SKU from March 2022.

December 3, 2021

A new image for Windows Server 2019

Version: 21.12.03

Primary changes:

  • Updated pytorch to version 1.10.0
  • Updated tensorflow to version 2.7.0
  • Fix for Azure Machine Learning SDK & AutoML environment
  • Windows Security update
  • Improvement of stability and minor bug fixes

November 4, 2021

A new image for Ubuntu 18.04

Version: 21.11.04

Primary changes:

  • Changed .NET Framework to version 3.1.414
  • Changed Azcopy to version 10.13.0
  • Changed Azure CLI to version 2.30.0
  • Changed CUDA to version 11.5
  • Changed Docker to version 20.10.10
  • Changed Intellijidea to version 2021.2.3
  • Changed NVIDIA Drivers to version 470.103.01
  • Changed NVIDIA SMI to version 470.103.01
  • Changed Nodejs to version v16.13.0
  • Changed Pycharm to version 2021.2.3
  • Changed VS Code to version 1.61.2
  • Conda
    • azureml_py36_automl
      • Changed azureml-core to version 1.35.0
    • py38_default
      • Changed Jupyter Lab / jupyterlab to version 3.2.1
      • Changed Jupyter Notebook / notebook to version 6.4.5
      • Changed Jupyter Server / jupyter_server to version 1.11.2
      • Changed PyTorch Profiler TensorBoard Plugin / torch-tb-profiler to version 0.3.1
      • Changed azure-core to version 1.19.1
      • Changed matplotlib to version 3.4.3
      • Changed mkl to version 2021.4.0
      • Changed onnx to version 1.10.2
      • Changed opencv-python to version 4.5.4.58
      • Changed pandas to version 1.3.4
      • Changed pytorch to version 1.10.0
      • Changed scikit-learn to version 1.0.1
      • Changed tensorflow-gpu to version 2.6.2

October 7, 2021

A new image for Ubuntu 18.04

Version: 21.10.07

Primary changes:

  • Changed pytorch to version 1.9.1
  • Changed Docker to version 20.10.9
  • Changed Intellijidea to version 2021.2.2
  • Changed Nodejs to version v14.18.0
  • Changed Pycharm to version 2021.2.2
  • Changed VS Code to version 1.60.2
  • Fixed AutoML environment (azureml_py36_automl)
  • Fixed Azure Storage Explorer stability
  • Improvement of stability and minor bug fixes

August 11, 2021

A new image for Windows Server 2019

Version: 21.08.11

Primary changes:

  • Windows Security update
  • Update of Nvidia CuDNN to 8.1.0
  • Update of Jupyter Lab -to 3.0.16
  • Added MLFLow for experiment tracking
  • Improvement of stability and minor bug fixes

July 12, 2021

A new image for Ubuntu 18.04

Primary changes:

  • Updated to PyTorch 1.9.0
  • Updated Azure CLI to 2.26.1
  • Updated Azure CLI Azure Machine Learning extension to 1.29.0
  • Update VS Code version 1.58.1
  • Improvement of stability and minor bug fixes

June 22, 2021

A new image for Windows Server 2019

Version: 21.06.22

Primary changes:

  • Updated to PyTorch 1.9.0
  • Fixed a bug where git wasn't available

June 1, 2021

A new image for Ubuntu 18.04

Version: 21.06.01

Primary changes:

  • Docker is enabled by default
  • JupyterHub uses JupyterLab by default
  • Updated Python versions to fix CVE-2020-15523
  • Updated IntelliJ IDEA to version 2021.1 to fix CVE-2021-25758
  • Updated PyCharm Community to 2021.1
  • Updated TensorFlow to version 2.5.0

Removed several icons from desktop.

May 22, 2021

A new image for Windows Server 2019

Version: 21.05.22

Selected version updates include:

  • AzCopy 10.10.0
  • Azure CLI 2.23.0
  • Azure Data Studio 1.28.0
  • CUDA 11.1
  • Java 11
  • Julia 1.0.5
  • Jupyter Lab 2.2.6
  • Microsoft Edge browser
  • NodeJS 16.2.0
  • Power BI Desktop 2.93.641.0 64-bit (May 2021)
  • PyCharm Community Edition 2021.1.1
  • Python 3.8
  • PyTorch 1.8.1
  • R 4.1.0
  • RStudio 1.4.1106
  • Spark 3.1.1
  • Storage Explorer 1.19.1
  • TensorFlow 2.5.0
  • Visual Studio Code 1.56.2 incl. Azure Machine Learning extension
  • Visual Studio Community Edition 2019 (version 16.9.6)

Removed Firefox, Apache Drill and Microsoft Integration Runtime.

Dark mode; changed icons on desktop; wallpaper background change.

May 12, 2021

A new image for Ubuntu 18.04

Selected version updates include:

  • azcopy 10.10
  • Azure CLI 2.23.0
  • Azure Data Studio 1.22.1
  • Azure Storage Explorer 1.19.1
  • CUDA 11.3, cuDNN 8, NCCL2
  • dask 2021.01.0
  • Java 11 (OpenJDK)
  • Jupyter Lab 3.0.14
  • Microsoft Edge browser (beta)
  • Python 3.8
  • PyTorch 1.8.1 incl. torchaudio torchtext torchvision, torch-tb-profiler
  • R 4.0.5
  • Spark 3.1 incl. mmlspark, connectors to Blob Storage, Data Lake, Azure Cosmos DB
  • TensorFlow 2.4.1 incl. TensorBoard
  • VS.Code 1.56

Added docker. To save resources, the docker service isn't started by default. To start the docker service, run this command at the command line:

sudo systemctl start docker

Note

If your machine has GPU(s), you can make use of the GPU(s) inside the containers by adding a --gpus parameter to your docker command.

For example, this command

sudo docker run --gpus all -it --rm -v local_dir:container_dir nvcr.io/nvidia/pytorch:18.04-py3

runs an Ubuntu 18.04 container with PyTorch pre-installed and all GPUs enabled. It also builds a local local_dir folder available in the container under container_dir.

February 24, 2020

The Data Science Virtual Machine (DSVM) images for Ubuntu 18.04 and Windows 2019 images are now available.

October 8, 2019

Updates to software on the Windows DSVM

  • Azure Storage Explorer 1.10.1
  • Firefox 69.0.2
  • Power BI Desktop 2.73.55xx
  • PyCharm 19.2.3
  • RStudio 1.2.50xx

Default Browser for Windows updated

Earlier, the default browser was set to Internet Explorer. Users are now prompted to choose a default browser when they first sign in.