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

Update softmax with parameterization #16

Merged
merged 12 commits into from
Apr 21, 2021
Merged

Update softmax with parameterization #16

merged 12 commits into from
Apr 21, 2021

Conversation

bdeonovic
Copy link
Contributor

parameterize softmax to be usable with AD.

@bdeonovic
Update basicfuns.jl
2900227 
parameterize softmax to be usable with AD.
devmotion
devmotion requested changes Apr 20, 2021
View reviewed changes
src/basicfuns.jl Outdated Show resolved Hide resolved
@devmotion
Copy link
Member

Depending on your use case, you might need to improve the implementation of softmax! as well (e.g., I guess <:AbstractFloat is quite restrictive and s = 0. should probably not be hardcoded as Float64).

@bdeonovic @devmotion
Update src/basicfuns.jl
dd5f78e 
incorporate suggestions from @devmotion

Co-authored-by: David Widmann <devmotion@users.noreply.github.com>
@bdeonovic
Copy link
Contributor Author

Do you have any comments on why softmax! was written the way it was? Is there a point to using inv rather than just dividing by the sum s?

@devmotion
Copy link
Member

Do you have any comments on why softmax! was written the way it was? Is there a point to using inv rather than just dividing by the sum s?

No, I don't know why it was written in this way. I assume that it uses inv for performance reasons: usually multiplication is faster than division.

devmotion
devmotion reviewed Apr 21, 2021
View reviewed changes
src/basicfuns.jl Outdated Show resolved Hide resolved
src/basicfuns.jl Outdated Show resolved Hide resolved
src/basicfuns.jl Show resolved Hide resolved
test/basicfuns.jl Show resolved Hide resolved
bdeonovic and others added 3 commits April 21, 2021 08:04
@bdeonovic @devmotion
Update src/basicfuns.jl
332d8f6 
Co-authored-by: David Widmann <devmotion@users.noreply.github.com>
@bdeonovic @devmotion
Update src/basicfuns.jl
61cd452 
Co-authored-by: David Widmann <devmotion@users.noreply.github.com>
@bdeonovic
Update basicfuns.jl
6caf987
devmotion
devmotion reviewed Apr 21, 2021
View reviewed changes
Copy link
Member

@devmotion devmotion left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

OK, looks good! If you revert back to using invs = inv(s), then it should be good to go and I'll merge and make a new release.

devmotion
devmotion reviewed Apr 21, 2021
View reviewed changes
src/basicfuns.jl Outdated Show resolved Hide resolved
@bdeonovic
Copy link
Contributor Author

thanks for your patience

devmotion
devmotion approved these changes Apr 21, 2021
View reviewed changes
Copy link
Member

@devmotion devmotion left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks a lot for the PR!

@devmotion devmotion merged commit 3289f12 into JuliaStats:master Apr 21, 2021
@devmotion
Copy link
Member

A new release with the fix should be available soon: JuliaRegistries/General#34861

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@devmotion devmotion devmotion approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@bdeonovic @devmotion