Skip to main content

More Flexibility in Worker Output Format

· One min read

Customers can now disable the automatic archiving of job outputs prior to upload.

How It Works​

Currently, this option is only available when creating jobs from the trainML SDK. Simply specify the archive=False field in the output_options dictionary for any output type. This option has no effect for trainml or regional output types.

job = await trainml.jobs.create(
    ...
    data=dict(
        ...
        output_type="aws",
        output_uri="s3://example-bucket/output",
        output_options=dict(archive=False),
    ),
    ...
)