-
Notifications
You must be signed in to change notification settings - Fork 28
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Generic Taskcluster task naming #589
Conversation
da0d6e4
to
f790c3f
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Approving but let's discuss if we want to implement support of renaming from #544 right away.
f790c3f
to
35133ed
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@vrigal Do I understand correctly that now a suffix -1
is added to all the runs? If yes, we do not want that. We train multiple models only for the teacher. This -1
makes names longer. To be more explicit, we want:
(for older models)
teacher-base-0
teacher-base-1
teacher-parallel-0
teacher-parallel-1
teacher-all-0
teacher-all-1
finetune-teacher-1
finetune-teacher-2
etc.
(for newer Taskcluster models)
teacher-1
teacher-2
backwards
student
student-finetuned
quantized
teacher-ensemble
Naming of the teacher models will differ for the older experiments but everything else should be exactly the same across all the experiments and publishers. We might want to store a map of renamings.
469781d
to
f17b963
Compare
2706f10
to
6c3c411
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ok, I see the values in unit tests but to verify this we would need to reupload everything from scratch. We can do that in #574
Required by #519 (making
build_task_name
supports evaluation tasks labels tags from Taskcluster)Closes #573