Asset naming convention

General

These conventions apply to all assets, regardless of type. Names must:

• Start with a lowercase letter.
• Use only lowercase letters, numbers, dashes, and periods.
• Be up to or below 255 characters long.
• Not include prefixes like "azureml" or "microsoft". Registries containing these assets will already provide this information.
• Not include the type of asset as a suffix (examples: "component" or "environment"). The asset type is already clearly represented via CLI, SDK, and UI.

Environments

Current environment naming convention

Your environment will need a meaningful name that will make it easy to find. The current naming convention is:

<framework>-<framework-version>-<os><os-version>-py<python-version>-{cpu|gpu}[-inference]

Field	Description	Examples
framework	Machine learning framework supported by the environment	pytorch, sklearn, tensorflow
framework-version	Framework version, major and minor values only	1.1, 11.1, 2.8
os	Operating system	ubuntu
os-version	Operating system version	20.04
python-version	Python version, major and minor values only, no period	38 (for Python 3.8)
cpu|gpu	Whether the environment includes GPU support, or CPU only	cpu, gpu
-inference	Suffix to indicate whether the environment supports AzureML inferencing use	-inference

Full examples:

• pytorch-1.11-ubuntu20.04-py38-cuda11-gpu
• sklearn-1.1-ubuntu20.04-py38-cpu
• tensorflow-2.8-ubuntu20.04-py38-cuda11-gpu

Simplified environment naming convention

The above naming convention may be simplified soon to something like this:

<framework>-<framework-version>[<gpu-driver><gpu-driver-version>-gpu]

Field	Description	Examples
framework	Machine learning framework supported by the environment	pytorch, sklearn, tensorflow
framework-version	Framework version, major and minor values only	1.1, 11.1, 2.8
gpu-driver	GPU driver	cuda
gpu-driver-version	GPU driver version, major and minor values only	11.6

Full examples:

• pytorch-1.9 (CPU version)
• pytorch-1.9-cuda11.6-gpu