Skip to content
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

Allow format_checkpoint_name() to use callback metrics #3254

Closed
minhduc0711 opened this issue Aug 29, 2020 · 1 comment · Fixed by #3598
Closed

Allow format_checkpoint_name() to use callback metrics #3254

minhduc0711 opened this issue Aug 29, 2020 · 1 comment · Fixed by #3598
Labels
feature Is an improvement or enhancement help wanted Open to be worked on

Comments

@minhduc0711
Copy link
Contributor

🚀 Feature

Let users use checkpoint_on and val_checkpoint_on in the checkpoint's filepath when initializing a ModelCheckpoint like this:

ckpt_callback = ModelCheckpoint(filepath="ckpts/{epoch}-{val_checkpoint_on}")

Motivation

Right now, the formatting function only allows selecting from metrics that were logged.

I would like to use metrics from trainer.callback_metrics as well. The reason I can't use logged metrics is that my scalar name contains a forward slash to utilize Tensorboard's grouping, and it would mess up the formatting function.

Pitch

Like what I said at the start.

Alternatives

Or perhaps we can adjust format_checkpoint_name() to handle forward slashes in metrics' names, but I guess that would be a little bit more tricky.

@minhduc0711 minhduc0711 added feature Is an improvement or enhancement help wanted Open to be worked on labels Aug 29, 2020
@github-actions
Copy link
Contributor

Hi! thanks for your contribution!, great first issue!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
feature Is an improvement or enhancement help wanted Open to be worked on
Projects
None yet
Development

Successfully merging a pull request may close this issue.

1 participant