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

Expose nclasses option for DenseNet #154

Merged
merged 2 commits into from
May 15, 2022

Conversation

theabhirath
Copy link
Member

Previously, the nclasses option was not exposed for the DenseNet API with the config option.

This PR adds that. Would be good to have a patch release too if possible (only with this commit and not the previous breaking change), since I'm trying to train some models and this will help with that 😅

@darsnack
Copy link
Member

Instead of maintaining a separate non-breaking branch, maybe the easiest thing would be to revert the breaking conv + bn changes since they weren't a major impact to TTFG anyways. We can release the other changes along with this and keep the breaking stuff in a separate, new PR.

@theabhirath
Copy link
Member Author

Instead of maintaining a separate non-breaking branch, maybe the easiest thing would be to revert the breaking conv + bn changes since they weren't a major impact to TTFG anyways. We can release the other changes along with this and keep the breaking stuff in a separate, new PR.

I've opened #155 to address this. I will run some more tests when I get time to ascertain just how much returning a Chain helps for conv_bn but for now, this should help make #151 non-breaking

Previously, the `nclasses` option was not exposed for the `DenseNet` API with the `config` option.
@darsnack
Copy link
Member

Can you bump the patch too?

@darsnack darsnack merged commit 2d67499 into FluxML:master May 15, 2022
@theabhirath theabhirath deleted the densenet-classes branch May 15, 2022 15:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants