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.

Sign up for GitHub

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

Jump to bottom

Checkpointer Saves Model when it outputs NaN #64

Open
sciapponi opened this issue Dec 5, 2023 · 1 comment
Open

Checkpointer Saves Model when it outputs NaN #64

sciapponi opened this issue Dec 5, 2023 · 1 comment
Assignees
@fpaissan
Labels
bug Something isn't working

Comments

@sciapponi
Copy link

When I'm training a model and gradients explode/vanish Micromind still saves it to the checkpoint folder.
Personally I'd like the framework to keep the last working version since it could be useful to identify issues about the weight computation.

image
@fpaissan
Copy link
Collaborator

fpaissan commented Dec 5, 2023

Thanks, @sciapponi, for letting us know about this problem.
I'll take a look an create a PR once this is tested. I'll make sure to keep you posted here.

@fpaissan fpaissan self-assigned this Dec 5, 2023
@fpaissan fpaissan added the bug Something isn't working label Dec 5, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@fpaissan fpaissan

Labels
bug Something isn't working
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@fpaissan @sciapponi