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

Support plates in RenyiELBO #2321

Merged
merged 1 commit into from
Feb 22, 2020
Merged

Support plates in RenyiELBO #2321

merged 1 commit into from
Feb 22, 2020

Conversation

fritzo
Copy link
Member

@fritzo fritzo commented Feb 21, 2020

Addresses #2220
Pair coded with @martinjankowiak

This adds plate support to RenyiELBO. Once this merges, similar changes can be applied to TraceAdaptiveTail_ELBO and ReweightedWakeSleep.

Tested

  • strengthened inference tests
  • added weak tests with various batch shapes
  • ... other test suggestions welcome...

@fritzo
Copy link
Member Author

fritzo commented Feb 21, 2020

cc @tuananhle7 @iffsid

@iffsid
Copy link
Contributor

iffsid commented Feb 21, 2020

cool, thanks!
We'll setup a PR once this merges. :)

@fehiepsi
Copy link
Member

The change looks reasonable to me. Just to verify my understanding, to support the previous behavior, I should declare different plates for z and x?

def model():
    with pyro.plate('z', 2):
        pyro.sample('z', ...)
    with pyro.plate('x', 2):
        pyro.sample('x', ...)

@fritzo
Copy link
Member Author

fritzo commented Feb 21, 2020

@fehiepsi yes good point, you can indeed declare different plates to achieve previous behavior.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants