ASF subversion and git services commented on AIRFLOW-2291:

Commit d9bf1edd44af847aa1f9ecd9d797dfacb806cc00 in incubator-airflow's branch 
refs/heads/master from [~dlebech]
[ https://git-wip-us.apache.org/repos/asf?p=incubator-airflow.git;h=d9bf1ed ]

[AIRFLOW-2291] Add optional params to ML Engine

This commit adds three extra optional parameters
to the
`MLEngineTrainingOperator` as well as a new unit
test for the

Closes #3202 from dlebech/ml-engine-python-version

> Ability to specify runtimeVersion, pythonVersion and jobDir in ML Engine 
> operator
> ---------------------------------------------------------------------------------
>                 Key: AIRFLOW-2291
>                 URL: https://issues.apache.org/jira/browse/AIRFLOW-2291
>             Project: Apache Airflow
>          Issue Type: Improvement
>          Components: contrib, gcp
>    Affects Versions: Airflow 2.0, Airflow 1.8, Airflow 1.9.0
>            Reporter: David Volquartz Lebech
>            Assignee: David Volquartz Lebech
>            Priority: Minor
> For the Google Cloud ML Engine Training Operator (in {{contrib}}), the 
> [current list of 
> arguments|https://github.com/apache/incubator-airflow/blob/b918a4976a641a3c944ef4ad4323fe25bd219ea0/airflow/contrib/operators/mlengine_operator.py#L538-L544]
>  does not seem to include the option for specifying runtime and Python 
> version which limits execution to Python 2.7. It also does not include the 
> useful job directory argument.
> It seems that simply adding {{runtimeVersion}}, {{pythonVersion}} and 
> {{jobDir}} to the list would be enough to make this work, according to [the 
> documentations on ML 
> engine|https://cloud.google.com/ml-engine/docs/tensorflow/versioning#set-python-version-training].

This message was sent by Atlassian JIRA

Reply via email to