This browser is no longer supported.

Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.

Prerequisites

Default install

Use azureml-core .

pip install azureml-core
Then install any other packages required for your particular job.
Upgrade install
We recommend that you always keep azureml-core updated to the latest version.
Upgrade a previous version:
pip install --upgrade azureml-core
Check version
Verify your SDK version:
pip show azureml-core
To see all packages in your environment:
pip list
You can also show the SDK version in Python, but this version does not include the minor version.
import azureml.core
print(azureml.core.VERSION)
To learn more about how to configure your development environment for Azure Machine Learning service, see Configure your development environment.
Other azureml packages
The SDK contains many other optional packages that you can install. These include dependencies that aren't required for all use-cases, so they are not included in the default installation in order to avoid bloating the environment. The following table outlines the packages ,their use-cases and command to install, update & version check.
Additional package 
Use-case
Install/Upgrade/Show version                
azureml-automl-core
Contains core automated machine learning classes for Azure Machine Learning.
 This package is used by azureml-train-automl-client and azureml-train-automl-runtime.
pip install azureml-automl-core 
  pip install --upgrade azureml-automl-core 
  pip show azureml-automl-core
azureml-accel-models
Accelerates deep neural networks on FPGAs with the Azure ML Hardware Accelerated Models Service.
pip install azureml-accel-models 
  pip install --upgrade azureml-accel-models 
  pip show azureml-accel-models
azureml-train-automl
Provides classes for building and running automated machine learning experiments. Also installs common data science packages including pandas, numpy, and scikit-learn. 

 If you're looking to submit automated ML runs on a remote compute and don't need do any ML locally, we recommend using the thin client, azureml-train-automl-client, package that is  part of the azureml-sdk.   
 
 See the additional use-case guidance for more information on installation and working with the full automl SDK or its thin client, azureml-train-automl-client.
 
 Similar to the Python standard, one version backwards and one version forward compatibility is supported, but only for the full azureml-train-automl package. For example, if a model is trained with SDK version 1.29.0, then you can inference with SDK versions between 1.28.0 and 1.30.0.
For local conda environment: 
 pip install azureml-train-automl 
  pip install --upgrade azureml-train-automl pip install show azureml-train-automl 

Thin client for remote compute: 
 pip install azureml-train-automl-client 
 pip install --upgrade azureml-train-automl-client 
 pip install show azureml-train-automl-client
azureml-contrib
Installs azureml-contrib-* packages, which include experimental functionality or preview features.
pip install azureml-contrib 
  pip install --upgrade azureml-contrib 
  pip show azureml-contrib
azureml-datadrift
Contains functionality to detect when model training data has drifted from its scoring data.
pip install azureml-datadrift 
  pip install --upgrade azureml-datadrift 
  pip show azureml-datadrift
azureml-interpret
Used for model interpretability, including feature and class importance for blackbox and whitebox models.
pip azureml-interpret 
  pip install --upgrade azureml-interpret 
  pip show azureml-interpret
azureml-widgets
Contains core packages, modules, and classes for Azure Machine Learning.
pip install azureml-widgets 
  pip install --upgrade azureml-widgets 
  pip show azureml-widgets
azureml-contrib-services
Provides functionality for scoring scripts to request raw HTTP access.
pip install azureml-contrib-services 
  pip install --upgrade azureml-contrib-services 
  pip show azureml-contrib-services
azureml-tensorboard
Provides classes and methods for exporting experiment run history and launching TensorBoard for visualizing experiment performance and structure.
pip install azureml-tensorboard 
  pip install --upgrade azureml-tensorboard 
  pip show azureml-tensorboard
azureml-mlflow
Contains functionality integrating Azure Machine Learning with MLFlow.
pip install azureml-mlflow 
  pip install --upgrade azureml-mlflow 
  pip show azureml-mlflow
azureml-automl-runtime
Contains automated machine learning classes for executing runs in Azure Machine Learning.
pip install azureml-automl-runtime 
  pip install --upgrade azureml-automl-runtime 
  pip show azureml-automl-runtime
azureml-widgets
Contains functionality to view the progress of machine learning training runs in Jupyter Notebooks.
pip install azureml-widgets 
  pip install --upgrade azureml-widgets 
  pip show azureml-widgets
azureml-train-restclients-hyperdrive
Contains classes needed to create HyperDriveRuns with azureml-train-core.
pip install azureml-train-restclients-hyperdrive 
  pip install --upgrade azureml-train-restclients-hyperdrive 
  pip show azureml-train-restclients-hyperdrive
azureml-train-core
Contains base estimator classes and the generic estimator class ,estimators used in Deep Neural Network (DNN) training ,estimators used in Scikit-Learn training, modules and classes supporting hyperparameter tuning.
pip install azureml-core 
  pip install --upgrade azureml-core 
  pip show azureml-core
azureml-train-automl-runtime
Contains functionality representing core automated ML and runtime components in Azure Machine Learning.
pip install azureml-train-automl-runtime 
  pip install --upgrade azureml-train-automl-runtime 
  pip show azureml-train-automl-runtime
azureml-train-automl-client
Contains core packages, modules, and classes for Azure Machine Learning.
pip install azureml-train-automl-client 
  pip install --upgrade azureml-train-automl-client 
  pip show azureml-train-automl-client
azureml-telemetry
This package is used to collect telemetry data like log messages, metrics, events, and activity messages.
pip install azureml-telemetry 
  pip install --upgrade azureml-telemetry 
  pip show azureml-telemetry
azureml-synapse
Contains Magic command to manage Synapse session and submit code and SparkMonitor widget to monitor the spark job progress, for both Jupyter and JupyterLab
pip install azureml-synapse 
  pip install --upgrade azureml-synapse 
  pip show azureml-synapse
azureml-sdk
Thos package is used to build and run machine learning workflows upon the Azure Machine Learning service
pip install azureml-sdk 
  pip install --upgrade azureml-sdk 
  pip show azureml-sdk
azureml-pipeline-steps
Contains pre-built steps that can be executed in an Azure Machine Learning Pipeline.
pip install azureml-pipeline-steps 
  pip install --upgrade azureml-pipeline-steps 
  pip show azureml-pipeline-steps
azureml-pipeline-core
Contains core functionality for Azure Machine Learning pipelines, which are configurable machine learning workflows.
pip install azureml-pipeline-core 
  pip install --upgrade azureml-pipeline-core 
  pip show azureml-pipeline-core
azureml-pipeline
This package is used to build, optimize, and manage  machine learning workflows
pip install azureml-pipeline 
  pip install --upgrade azureml-pipeline 
  pip show azureml-pipeline
azureml-opendatasets
Contains core functionality for Azure Machine Learning pipelines, which are configurable machine learning workflows.
pip install azureml-opendatasets 
  pip install --upgrade azureml-opendatasets 
  pip show azureml-opendatasets
azureml-interpret
Contains functionality for working with model interpretability in Azure Machine Learning.
pip install azureml-interpret 
  pip install --upgrade azureml-interpret 
  pip show azureml-interpret
azureml-defaults
This package is a metapackage that is used internally by Azure Machine Learning.
pip install azureml-defaults 
  pip install --upgrade azureml-defaults 
  pip show azureml-defaults
azureml-dataset-runtime
The purpose of this package is to coordinate dependencies within AzureML packages. This package is internal, and is not intended to be used directly.
pip install azureml-dataset-runtime 
  pip install --upgrade azureml-dataset-runtime 
  pip show azureml-dataset-runtime
azureml-datadrift
Contains functionality to detect when model training data has drifted from its scoring data.
pip install azureml-datadrift 
  pip install --upgrade azureml-datadrift 
  pip show azureml-datadrift
azureml-contrib-server
This package is local HTTP service used to expose a subset of the functionality provided by the AzureML SDK to VS Tools for AI extensions (VSCode and Visual Studio)
pip install azureml-contrib-server 
  pip install --upgrade azureml-contrib-server 
  pip show azureml-contrib-server
azureml-contrib-run
This package is used to contain the integration code of AzureML with Mlflow.
pip install azureml-core 
  pip install --upgrade azureml-core 
  pip show azureml-core
azureml-contrib-reinforcementlearning
Contains functionality for creating a Windows compute target in Azure Machine Learning.
pip install azureml-contrib-reinforcementlearning 
  pip install --upgrade azureml-contrib-reinforcementlearning 
  pip show azureml-contrib-reinforcementlearning
azureml-contrib-pipeline-steps
Contains modules and classes for specialized Azure Machine Learning Pipeline steps and associated configuration.
pip install azureml-contrib-pipeline-steps 
  pip install --upgrade azureml-contrib-pipeline-steps 
  pip show azureml-contrib-pipeline-steps
azureml-contrib-notebook
Contains extensions for working with Jupyter notebooks in Azure Machine Learning.
pip install azureml-contrib-notebook 
  pip install --upgrade azureml-contrib-notebook 
  pip show azureml-contrib-notebook
azureml-contrib-gbdt
This package contains LightGBM estimator.
pip install azureml-contrib-gbdt 
  pip install --upgrade azureml-contrib-gbdt 
  pip show azureml-contrib-gbdt
azureml-contrib-functions
Contains functionality for packaging Azure Machine Learning models for deployment to Azure Functions.
pip install azureml-contrib-functions 
  pip install --upgrade azureml-contrib-functions 
  pip show azureml-contrib-functions
azureml-contrib-fairness
This package supports the use of fairness assessment dashboards in Azure Machine Learning Studio
pip install azureml-contrib-fairness 
  pip install --upgrade azureml-contrib-fairness 
  pip show azureml-contrib-fairness
azureml-contrib-dataset
Contains specialized functionality for working with Dataset objects in Azure Machine Learning.
pip install azureml-contrib-dataset 
  pip install --upgrade azureml-contrib-dataset 
  pip show azureml-contrib-dataset
azureml-contrib-automl-pipeline-steps
Contains pre-built steps that can be executed in an Azure Machine Learning Pipeline.
pip install azureml-contrib-automl-pipeline-steps 
  pip install --upgrade azureml-contrib-automl-pipeline-steps 
  pip show azureml-contrib-automl-pipeline-steps
azureml-contrib-automl-dnn-vision
This package is only meant to be used by AutoML system-generated scripts. To install in Windows, the “torch” and “torchvision” packages must be installed separately before this package.
pip install azureml-contrib-automl-dnn-vision 
  pip install --upgrade azureml-contrib-automl-dnn-vision 
  pip show azureml-contrib-automl-dnn-vision
azureml-contrib-automl-dnn-forecasting
Azure ML CLI extension common package. Common across azure-cli-ml and azure-cli-ml-preview.
pip install azureml-contrib-automl-dnn-forecasting 
  pip install --upgrade azureml-contrib-automl-dnn-forecasting 
  pip show azureml-contrib-automl-dnn-forecasting
azureml-contrib-aisc
AzureML Contrib for AzureML AI Super Computer compute target. AISCCompute is a managed AI compute infrastructure, which can be attached to a workspace by cluster admin.
pip install azureml-contrib-aisc 
  pip install --upgrade azureml-contrib-aisc 
  pip show azureml-contrib-aisc
azureml-cli-common
Azure ML CLI extension common package. Common across azure-cli-ml and azure-cli-ml-preview.
pip install azureml-cli-common 
  pip install --upgrade azureml-cli-common 
  pip show azureml-cli-common
azureml-automl-dnn-nlp
This package is only meant to be used by AutoML system-generated scripts.
pip install azureml-automl-dnn-nlp 
  pip install --upgrade azureml-automl-dnn-nlp 
  pip show azureml-automl-dnn-nlp
azureml-accel-models
Accelerate deep neural networks on FPGAs with the Azure ML Hardware Accelerated Models Service.
pip install azureml-accel-models 
  pip install --upgrade azureml-accel-models 
  pip show azureml-accel-models
azureml-inference-server-http
This package enable  Local Development,CI/CD Integration,Server Routes.
pip install azureml-inference-server-http 
  pip install --upgrade azureml-inference-server-http 
  pip show azureml-inference-server-http
azure-ml-component
This package contains functionality for authoring and managing Azure Machine Learning components  authoring and submiting pipelines using components
pip install azure-ml-component 
  pip install --upgrade azure-ml-component 
  pip show azure-ml-component
azureml-pipeline-wrapper
This package contains functionality for  authoring and managing Azure Machine Learning modules , authoring and submiting pipelines using modules
pip install azureml-pipeline-wrapper 
  pip install --upgrade azureml-pipeline-wrapper 
  pip show azureml-pipeline-wrapper
azureml-designer-cv-modules
Modules to pre-process and transform images such as to crop, pad or resize.
pip install azureml-designer-cv-modules 
  pip install --upgrade azureml-designer-cv-modules 
  pip show azureml-designer-cv-modules
azureml-designer-pytorch-modules
Modules to train and inference image classification models based on pytorch framework.
pip install azureml-designer-pytorch-modules 
  pip install --upgrade azureml-designer-pytorch-modules 
  pip show azureml-designer-pytorch-modules
azureml-designer-vowpal-wabbit-modules
Modules to train and inference models based on Vowpal Wabbit framework.
pip install azureml-designer-vowpal-wabbit-modules 
  pip install --upgrade azureml-designer-vowpal-wabbit-modules 
  pip show azureml-designer-vowpal-wabbit-modules
azureml-designer-classic-modules
A variety of modules for data processing, model training, inferencing and evaluation.
pip install azureml-designer-classic-modules 
  pip install --upgrade azureml-designer-classic-modules 
  pip show azureml-designer-classic-modules
azureml-designer-recommender-modules
Modules to train and inference recommendation models based on deep neural network.
pip install azureml-designer-recommender-modules 
  pip install --upgrade azureml-designer-recommender-modules 
  pip show azureml-designer-recommender-modules
azureml-designer-internal
Internal functionalities provided for built-in modules.
pip install azureml-designer-internal 
  pip install --upgrade azureml-designer-internal 
  pip show azureml-designer-internal
azureml-designer-core
Core functionalities for data-type definition, data io and frequently-used functions.
pip install azureml-designer-core 
  pip install --upgrade azureml-designer-core 
  pip show azureml-designer-core
azureml-designer-datatransform-modules
Modules to transform dataset, such as by applying math operations, sql queries, clipping outliers or generating a statistics report.
pip install azureml-designer-datatransform-modules 
  pip install --upgrade azureml-designer-datatransform-modules 
  pip show azureml-designer-datatransform-modules
azureml-designer-dataio-modules
Modules to load data into azure machine learning designer and write data to cloud-based storage.
pip install azureml-designer-dataio-modules 
  pip install --upgrade azureml-designer-dataio-modules 
  pip show azureml-designer-dataio-modules
azureml-designer-serving
Provide functionalities to invoke built-in modules in deployment service.
pip install azureml-designer-serving 
  pip install --upgrade azureml-designer-serving 
  pip show azureml-designer-serving
azureml-contrib-datadrift
Contains functionality for data drift detection for various datasets used in machine learning, including training datasets and scoring dataset.
pip install azureml-contrib-datadrift 
  pip install --upgrade azureml-contrib-datadrift 
  pip show azureml-contrib-datadrift
azureml-contrib-explain-model
Contains experimental functionality for the azureml-explain-model package, which offers a variety of services for machine learning model interpretability.
pip install azureml-contrib-explain-model 
  pip install --upgrade azureml-contrib-explain-model 
  pip show azureml-contrib-explain-model
azureml-contrib-opendatasets
This package provides a set of APIs to consume Azure Open Datasets.
pip install azureml-contrib-opendatasets 
  pip install --upgrade azureml-contrib-opendatasets 
  pip show azureml-contrib-opendatasets
azureml-train-widgets
Contains widgets for Jupyter Notebooks to visually track your runs.
pip install azureml-train-widgets 
  pip install --upgrade azureml-train-widgets 
  pip show azureml-train-widgets
For more details on above packages, see AzureML on pypi.
Additional use-case guidance
If your use-case is described below, note the guidance and any recommended actions.
Use-case
Guidance
Using automl 
Install the full azureml-train-automl SDK in a new 64-bit Python environment. A new 64-bit environment is required because of a dependency on the LightGBM framework. This package installs and pins specific versions of data science packages for compatibility, which requires a clean environment. 

 The thin client, azureml-train-automl-client, package doesn't install additional data science packages or require a clean Python environment. We recommend azureml-train-automl-client  if you only need to submit automated ML runs to a remote compute, and don't need to submit local runs or download your model locally. 
 
 One version backwards and one version forward compatibility is only supported for models trained with the full azureml-train-automl package.  For example, if a model is trained with SDK version 1.29.0, then you can inference with SDK versions between 1.28.0 and 1.30.0.
Using Azure Databricks
In the Azure Databricks environment, use the library sources detailed in this guide for installing the SDK. Also, see these tips for further information on working with Azure Machine Learning SDK for Python on Azure Databricks.
Using Azure Data Science Virtual Machine
Azure Data Science Virtual Machines created after September 27, 2018 come with the Python SDK preinstalled.
Running Azure Machine Learning tutorials or notebooks
If you are using an older version of the SDK than the one mentioned in the tutorial or notebook, you should upgrade your SDK. Some functionality in the tutorials and notebooks may require additional Python packages such as matplotlib, scikit-learn, or pandas. Instructions in each tutorial and notebook will show you which packages are required.
Troubleshooting
Pip Installation: Dependencies are not guaranteed to be consistent with single-line installation:
This is a known limitation of pip, as it does not have a functioning dependency resolver when you install as a single line. The first  unique dependency is the only one it looks at.
In the following code azureml-datadrift and azureml-train-automl are both installed using a single-line pip install.
  pip install azureml-datadrift, azureml-train-automl
For this example, let's say azureml-datadrift requires version > 1.0 and azureml-train-automl requires version < 1.2. If the latest version of azureml-datadrift is 1.3,  then both packages get upgraded to 1.3, regardless of the azureml-train-automl package requirement for an older version.
To ensure the appropriate versions are installed for your packages, install using multiple lines like in the following code. Order isn't an issue here, since pip explicitly downgrades as part of the next line call. And so, the appropriate version dependencies are applied.
   pip install azureml-datadrift
   pip install azureml-train-automl 
Explanation package not guaranteed to be installed when installing the azureml-train-automl-client:
When running a remote AutoML run with model explanation enabled, you will see an error message "Please install azureml-explain-model package for model explanations." This is a known issue. As a workaround follow one of the steps below:
Install azureml-explain-model locally.
    pip install azureml-explain-model
Disable the explainability feature entirely by passing model_explainability=False in the AutoML configuration.
    automl_config = AutoMLConfig(task = 'classification',
                           path = '.',
                           debug_log = 'automated_ml_errors.log',
                           compute_target = compute_target,
                           run_configuration = aml_run_config,
                           featurization = 'auto',
                           model_explainability=False,
                           training_data = prepped_data,
                           label_column_name = 'Survived',
                           **automl_settings)
Panda errors: Typically seen during AutoML Experiment:
When manually setting up your environment using pip, you may notice errors (especially from pandas) due to unsupported package versions being installed.
For example, ModuleNotFoundError: No module named 'pandas.core.internals.managers'; 'pandas.core.internals' is not a package
In order to prevent such errors, please install the AutoML SDK using the automl_setup.cmd:
Open an Anaconda prompt and clone the GitHub repository for a set of sample notebooks.
git clone https://github.com/Azure/MachineLearningNotebooks.git
cd to the how-to-use-azureml/automated-machine-learning folder where the sample notebooks were extracted and then run:
automl_setup
KeyError: 'brand' when running AutoML on local compute or Azure Databricks cluster
If a new environment was created after June 10, 2020, by using SDK 1.7.0 or earlier, training might fail with this error due to an update in the py-cpuinfo package. (Environments created on or before June 10, 2020, are unaffected, as are experiments run on remote compute because cached training images are used.) To work around this issue, take either of the following two steps:
Update the SDK version to 1.8.0 or later (this also downgrades py-cpuinfo to 5.0.0):
pip install --upgrade azureml-sdk[automl]
Downgrade the installed version of py-cpuinfo to 5.0.0:
pip install py-cpuinfo==5.0.0
Error message: Cannot uninstall 'PyYAML'
Azure Machine Learning SDK for Python: PyYAML is a distutils installed project. Therefore, we cannot accurately determine which files belong to it if there is a partial uninstall. To continue installing the SDK while ignoring this error, use:
pip install --upgrade azureml-sdk[notebooks,automl] --ignore-installed PyYAML
Azure Machine Learning SDK installation failing with an exception: ModuleNotFoundError: No module named 'ruamel' or 'ImportError: No module named ruamel.yaml'
This issue is getting encountered with the installation of Azure Machine Learning SDK for Python on the latest pip (>20.1.1) in the conda base environment for all released versions of Azure Machine Learning SDK for Python. Refer to the following workarounds:
Avoid installing Python SDK on the conda base environment, rather create your conda environment and install SDK on that newly created user environment. The latest pip should work on that new conda environment.
For creating images in docker, where you cannot switch away from conda base environment, please pin pip<=20.1.1 in the docker file.
conda install -c r -y conda python=3.8 pip=20.1.1
Next steps
Try these next steps to learn how to use the Azure Machine Learning service SDK for Python:
Read the Azure Machine Learnin Python SDK overview to learn about key classes and design patterns with code samples.
Follow Azure Machine Learning Python Get Started tutorial to begin creating experiments and models.