Pipelines: Is it possible to specify memory to an Op when using the func_to_container_op?

Created on 1 Feb 2020  路  1Comment  路  Source: kubeflow/pipelines

Hi

I'm wondering if it's possible to add set_memory_request or set_memory_limit to a Op that is created from func_to_container_op's factory.

Thanks

aresddsl kinquestion statutriaged

Most helpful comment

I'm wondering if it's possible to add set_memory_request or set_memory_limit to a Op that is created from func_to_container_op's factory.

Yes, you do it the same way (as the object you get is the same - ContainerOp instance).

You should set the memory request on the task object which you get when you call a component with some arguments.

train_task = train_classifier(training_data=....).set_memory_request(...)

>All comments

I'm wondering if it's possible to add set_memory_request or set_memory_limit to a Op that is created from func_to_container_op's factory.

Yes, you do it the same way (as the object you get is the same - ContainerOp instance).

You should set the memory request on the task object which you get when you call a component with some arguments.

train_task = train_classifier(training_data=....).set_memory_request(...)
Was this page helpful?
0 / 5 - 0 ratings

Related issues

Bobgy picture Bobgy  路  3Comments

discordianfish picture discordianfish  路  4Comments

zijianjoy picture zijianjoy  路  5Comments

VindhyaSRajan picture VindhyaSRajan  路  3Comments

talhairfanbentley picture talhairfanbentley  路  5Comments