az ml run

Commands for submitting, updating, and monitoring runs.

Commands

az ml run cancel

Cancel run.

az ml run download-logs

Download log files.

az ml run list

List runs.

az ml run monitor-logs

Monitor the logs for an existing run.

az ml run monitor-tensorboard

Monitor an existing run using tensorboard.

az ml run show

Show run.

az ml run submit-hyperdrive

Submit a hyper parameter sweep using run config.

az ml run submit-pipeline

Submit a pipeline for execution, from a published pipeline ID or pipeline YAML file.

az ml run submit-script

Submit a script for execution.

az ml run update

Update the run by adding tags.

az ml run cancel

Cancel run.

az ml run cancel --run
                 [--experiment-name]
                 [--output-metadata-file]
                 [--path]
                 [--resource-group]
                 [--subscription-id]
                 [--workspace-name]

Required Parameters

--run -r

The runId of an experiment run.

Optional Parameters

--experiment-name -e

Experiment name.

--output-metadata-file -t

Provide an optional output file location for structured object output.

--path

Path to a root directory for run configuration files.

default value: .
--resource-group -g

Resource group name.

--subscription-id

Specifies the subscription Id.

--workspace-name -w

Workspace name.

az ml run download-logs

Download log files.

az ml run download-logs --output-dir
                        --run
                        [--experiment-name]
                        [--output-metadata-file]
                        [--path]
                        [--resource-group]
                        [--subscription-id]
                        [--workspace-name]

Required Parameters

--output-dir -d

Output directory to download log files to.

--run -r

The runId of an experiment run.

Optional Parameters

--experiment-name -e

Experiment name.

--output-metadata-file -t

Provide an optional output file location for structured object output.

--path

Path to a root directory for run configuration files.

default value: .
--resource-group -g

Resource group name.

--subscription-id

Specifies the subscription Id.

--workspace-name -w

Workspace name.

az ml run list

List runs.

az ml run list [--compute-target-name]
               [--experiment-name]
               [--last]
               [--minimal]
               [--output-metadata-file]
               [--parent-run-id]
               [--path]
               [--pipeline-run-id]
               [--resource-group]
               [--status {Completed, Failed, Finalizing, Preparing, Provisioning, Queued, Running, Starting}]
               [--subscription-id]
               [--tags]
               [--workspace-name]

Optional Parameters

--compute-target-name

The compute target name.

--experiment-name -e

Experiment name.

--last

Fetch the latest N elements.

default value: 10
--minimal

Flag to provide minimum properties for run output.

--output-metadata-file -t

Provide an optional output file location for structured object output.

--parent-run-id

Parent Run ID.

--path

Path to a root directory for run configuration files.

default value: .
--pipeline-run-id

Pipeline Run ID.

--resource-group -g

Resource group name.

--status

Status of the run.

accepted values: Completed, Failed, Finalizing, Preparing, Provisioning, Queued, Running, Starting
--subscription-id

Specifies the subscription Id.

--tags

Tags for a run with 'key[=value]' syntax.

--workspace-name -w

Workspace name.

az ml run monitor-logs

Monitor the logs for an existing run.

az ml run monitor-logs --run
                       [--experiment-name]
                       [--output-metadata-file]
                       [--path]
                       [--resource-group]
                       [--subscription-id]
                       [--workspace-name]

Required Parameters

--run -r

The runId of an experiment run.

Optional Parameters

--experiment-name -e

Experiment name.

--output-metadata-file -t

Provide an optional output file location for structured object output.

--path

Path to a root directory for run configuration files.

default value: .
--resource-group -g

Resource group name.

--subscription-id

Specifies the subscription Id.

--workspace-name -w

Workspace name.

az ml run monitor-tensorboard

Monitor an existing run using tensorboard.

az ml run monitor-tensorboard --run
                              [--experiment-name]
                              [--local-directory]
                              [--output-metadata-file]
                              [--path]
                              [--resource-group]
                              [--subscription-id]
                              [--workspace-name]

Required Parameters

--run -r

The runId of an experiment run.

Optional Parameters

--experiment-name -e

Experiment name.

--local-directory

Local Directory to stage tensorboard files being monitored.

--output-metadata-file -t

Provide an optional output file location for structured object output.

--path

Path to a root directory for run configuration files.

default value: .
--resource-group -g

Resource group name.

--subscription-id

Specifies the subscription Id.

--workspace-name -w

Workspace name.

az ml run show

Show run.

az ml run show --run
               [--experiment-name]
               [--output-metadata-file]
               [--path]
               [--resource-group]
               [--subscription-id]
               [--workspace-name]

Required Parameters

--run -r

The runId of an experiment run.

Optional Parameters

--experiment-name -e

Experiment name.

--output-metadata-file -t

Provide an optional output file location for structured object output.

--path

Path to a root directory for run configuration files.

default value: .
--resource-group -g

Resource group name.

--subscription-id

Specifies the subscription Id.

--workspace-name -w

Workspace name.

az ml run submit-hyperdrive

Submit a hyper parameter sweep using run config.

az ml run submit-hyperdrive --hyperdrive-configuration-name
                            [--async]
                            [--conda-dependencies]
                            [--ct]
                            [--experiment-name]
                            [--output-metadata-file]
                            [--path]
                            [--resource-group]
                            [--run-configuration-name]
                            [--source-directory]
                            [--subscription-id]
                            [--workspace-name]
                            [<USER_SCRIPT_AND_ARGUMENTS>]

Required Parameters

--hyperdrive-configuration-name

The full name of the hyperdrive configuration file. The file should be in a sub-folder of the directory specified by the path parameter.

Optional Parameters

--async

Disable output streaming.

--conda-dependencies -d

Override the default Conda dependencies file.

--ct --target

The name of the compute target to use for the run.

--experiment-name -e

Experiment name.

--output-metadata-file -t

Provide an optional output file location for structured object output.

--path

Path to a root directory for run configuration files.

default value: .
--resource-group -g

Resource group name.

--run-configuration-name -c

Name (without extension) of a run configuration file. The file should be in a sub-folder of the directory specified by the path parameter.

--source-directory

A local directory containing source code files. Defaults to path if source directory is not provided.

--subscription-id

Specifies the subscription Id.

--workspace-name -w

Workspace name.

<USER_SCRIPT_AND_ARGUMENTS>

The run submit arguments, like script name and script arguments.

az ml run submit-pipeline

Submit a pipeline for execution, from a published pipeline ID or pipeline YAML file.

az ml run submit-pipeline [--datapaths]
                          [--experiment-name]
                          [--output-metadata-file]
                          [--output_file]
                          [--parameters]
                          [--path]
                          [--pipeline-id]
                          [--pipeline-yaml]
                          [--resource-group]
                          [--subscription-id]
                          [--workspace-name]

Optional Parameters

--datapaths -d

Comma-separated list of name=datastore/path pairs for datapath parameter assignments.

--experiment-name -n

Experiment name for run submission. If unspecified, use the pipeline name.

--output-metadata-file -t

Provide an optional output file location for structured object output.

--output_file -f

File to write output in JSON format.

--parameters -p

Comma-separated list of name=value pairs for pipeline parameter assignments.

--path

Path to a root directory for run configuration files.

default value: .
--pipeline-id -i

ID of a pipeline to submit (guid).

--pipeline-yaml -y

YAML file which defines a pipeline.

--resource-group -g

Resource group name.

--subscription-id

Specifies the subscription Id.

--workspace-name -w

Workspace name.

az ml run submit-script

Submit a script for execution.

az ml run submit-script [--async]
                        [--conda-dependencies]
                        [--ct]
                        [--experiment-name]
                        [--output-metadata-file]
                        [--path]
                        [--resource-group]
                        [--run-configuration-name]
                        [--source-directory]
                        [--subscription-id]
                        [--workspace-name]
                        [<USER_SCRIPT_AND_ARGUMENTS>]

Optional Parameters

--async

Disable output streaming.

--conda-dependencies -d

Override the default Conda dependencies file.

--ct --target

The name of the compute target to use for the run.

--experiment-name -e

Experiment name.

--output-metadata-file -t

Provide an optional output file location for structured object output.

--path

Path to a root directory for run configuration files.

default value: .
--resource-group -g

Resource group name.

--run-configuration-name -c

Name (without extension) of a run configuration file. The file should be in a sub-folder of the directory specified by the path parameter.

--source-directory

A local directory containing source code files. Defaults to path if source directory is not provided.

--subscription-id

Specifies the subscription Id.

--workspace-name -w

Workspace name.

<USER_SCRIPT_AND_ARGUMENTS>

The run submit arguments, like script name and script arguments.

az ml run update

Update the run by adding tags.

az ml run update --run
                 [--add-tag]
                 [--experiment-name]
                 [--output-metadata-file]
                 [--path]
                 [--resource-group]
                 [--subscription-id]
                 [--workspace-name]

Required Parameters

--run -r

The runId of an experiment run.

Optional Parameters

--add-tag

Tag the entitiy with 'key[=value]' syntax.

--experiment-name -e

Experiment name.

--output-metadata-file -t

Provide an optional output file location for structured object output.

--path

Path to a root directory for run configuration files.

default value: .
--resource-group -g

Resource group name.

--subscription-id

Specifies the subscription Id.

--workspace-name -w

Workspace name.