Job Labels

AI Platform provides you with the ability to label your jobs with key-value pairs. Any arguments you provide using either custom script arguments or an experiment broadcast will be added to your job as labels, like this:

In addition to arguments Caliban will add these labels to each job:

  • job_name: caliban_totoro by default, or the argument you pass using caliban cloud --name custom_name

  • gpu_enabled: true by default, or false if you ran your job with --nogpu

Cloud has fairly strict requirements on the format of each label’s key and value; Caliban will transform your arguments into labels with the proper formatting, so you don’t have to think about these.

Additional Custom Labels

You can also pass extra custom labels using -l or --label:

caliban cloud -l key:value --label another_k:my_value ...

These labels will be applied to every job if you’re running an experiment broadcast, or to the single job you’re submitting otherwise.

If you provide a label that conflicts with a user argument or experiment flag, your label will get knocked out.

Note

periods aren’t allowed in labels, but are often quite meaningful; because of this caliban replaces periods with underscores before stripping out any restricted characters.