-
Notifications
You must be signed in to change notification settings - Fork 512
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
1 changed file
with
1 addition
and
59 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,59 +1 @@ | ||
| # Curated Environments tied to SDK release | ||
|
|
||
| ## Authors' Guide | ||
|
|
||
| Curated environments let users to get started with Azure ML quickly and easily. They contain ML libraries relevant for specific scenario, with Microsoft stamp of approval. | ||
|
|
||
| Publishing a curated environment is equivalent to shipping a GA product. Once published, the environment cannot be pulled back without potentially breaking users' applications. As an author of curated environment, you: | ||
|
|
||
| * Commit to supporting the environment for the life of Azure ML service. | ||
| * Address bugs, customer issues and live site issues promptly. | ||
| * Ensure that environment and its documentation stay up-to date. | ||
| * Ensure you have CELA clearance to ship the environment and its dependencies. | ||
| * If leaving the team, transfer the environment to new owner. | ||
|
|
||
|
|
||
| ### Checklist | ||
|
|
||
| * There must be a valid use case for publishing a new environment | ||
| * First review if an existing environment could be adapted or updated. | ||
| * Do not publish a new environment just to advertise a new feature. | ||
| * Environment must have documentation such as Notebooks to show the use case | ||
| * Environment must be named "AzureML-\<Short Name\>" | ||
| * External packages must be high quality and actively maintained. | ||
| * External packages must support the OS and Python versions that Azure ML supports. | ||
| * Environment must build correctly as Docker Image. | ||
| * Make environment as small as possible and avoid unnecessary dependencies. Smaller environments build and load faster, resulting in better user experience. | ||
| * Avoid pinned dependencies. Pinned dependencies make it harder for user to customize the environment. | ||
|
|
||
| ### Process | ||
| * Create folder "Short Name" and place envdef.json and .contact files there. | ||
| * Start a PR with "curated-env-reviewers" as required reviewers. | ||
| * Environments shipped as part of AzureML-SDK release. | ||
| * Environment that fails to build as a Docker image will not ship. | ||
| * Environments are shipped only if there's been an update. | ||
| * Environments are versioned. By default, users will receive the latest version. | ||
| * Environments are distributed to: | ||
| * Global cache | ||
| * Data Science Instance (TBD) | ||
| * Once shipped, environments are visible through: | ||
| * SDK and CLI list queries | ||
| * Workspace UI (TBD) | ||
| * As Jupyter Kernels on DSI (TBD) | ||
|
|
||
|
|
||
| ## List of environments | ||
|
|
||
| | Environment | Use case | | ||
| |-----------------|---------------| | ||
| | AzureML-Minimal | Minimal environment that supports core Azure ML operations. Users can layer their custom packages on top.| | ||
| | AzureML-Tutorial | Environment intended for running tutorial Notebooks. Contains different Azure ML packages, and common data science packages, but no deep learning libraries.| | ||
| | AzureML-TensorFlow-\<tf-version\>-CPU | Environment for Tensorflow Estimator | | ||
| | AzureML-TensorFlow-\<tf-version\>-GPU | Environment for Tensorflow Estimator | | ||
| | AzureML-PyTorch-\<pytorch-version\>-CPU | Environment for PyTorch Estimator | | ||
| | AzureML-PyTorch-\<pytorch-version\>-GPU | Environment for PyTorch Estimator | | ||
| | AzureML-Chainer-\<chainer-version\>-CPU | Environment for Chainer Estimator | | ||
| | AzureML-Chainer-\<chainer-version\>-GPU | Environment for Chainer Estimator | | ||
| | AzureML-Scikit-Learn | Environment for Scikit-Learn Estimator | | ||
| | AzureML-AutoML | Environment for AutoML runs | ||
|
|
||
| # Azure ML Environment Examples |