issues with passing command line args to custom training job vertex AI

Viewed 22

I am trying to run basic custom training job

job = aiplatform.CustomContainerTrainingJob(
 display_name='testjob-name',
 container_uri='gcr.io/prj-id/image-name:latest',
 project=project_id,
 credentials= credentials,
 staging_bucket= 'stage-bucket'
)

using below code to run the job

job.run(
        args=['--data_dir', '/gcs/bucket/folder',
        '--model_dir', '/gcs/bucket/model',
        '--configs', 'internal-config.yml'],
        replica_count=1,
        sync=True
    )

Training job is getting exited with code 127 when I am passing args with job.run(). kindly help, what is the correct way to send command line arguments to custom training python script. There isn’t much coming up in logs either.

Thanks in advance

0 Answers
Related