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

Decouple updating t from update_stats #4543

Open
jackgerrits opened this issue Mar 31, 2023 · 0 comments
Open

Decouple updating t from update_stats #4543

jackgerrits opened this issue Mar 31, 2023 · 0 comments
Labels
Technical debt Things that are mis-done and need some fixing

Comments

@jackgerrits
Copy link
Member

The t value used to decay learning rate among other things is updated during update_stats. It would be better if t was updated automatically during learn instead of in a "finalization" method. That way update_stats could he be avoided entirely if loss is not needed.

We would need to update the semantics of t to only update during learn and then change this call to remove holdout and unlabeled references.

float t = static_cast<float>(

Confirmation on if updating t during only learn makes sense, but it seems to me to be valid.

We should also consider moving t outside of shared_data since all other fields are reporting related only.

@bassmang bassmang added the Technical debt Things that are mis-done and need some fixing label Apr 6, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Technical debt Things that are mis-done and need some fixing
Projects
None yet
Development

No branches or pull requests

2 participants