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

Softplus transform for AutoNormal scales [feature request] #2766

Closed
vitkl opened this issue Feb 18, 2021 · 1 comment · Fixed by #2767
Closed

Softplus transform for AutoNormal scales [feature request] #2766

vitkl opened this issue Feb 18, 2021 · 1 comment · Fixed by #2767
Assignees
@fritzo
Labels
enhancement

Comments

@vitkl
Copy link
Contributor

vitkl commented Feb 18, 2021 •
edited by fritzo
Loading

Hi @fritzo @fehiepsi

Following the conversation in numpyro issues (pyro-ppl/numpyro#855 (comment)) about the usefulness of softplus transform for AutoNormal scales, adding this feature request for further discussion here.

Requires pytorch/pytorch#52300
@fritzo fritzo self-assigned this Feb 19, 2021
@fritzo
Copy link
Member

fritzo commented Feb 19, 2021 •
edited
Loading

Thanks @vitkl! Mind if I sketch a potential solution for discussion?

EDIT I've added #2767 which accomplishes point 2 but not point 1 in pyro-ppl/numpyro#855 (comment)

This was referenced Feb 19, 2021
Merged
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@fritzo fritzo

Labels
enhancement
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Stabilize autoguide scale parameters via SoftplusTransform pyro-ppl/pyro
3 participants
@fritzo @fehiepsi @vitkl