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 energy for bfmi diagnostic #369

Merged
merged 2 commits into from
Sep 30, 2019
Merged

Conversation

fehiepsi
Copy link
Member

@fehiepsi fehiepsi commented Sep 29, 2019

Resolves #349.

This PR exposes energy, which is required for bfmi diagnostic, and energy plot. Because this is just a scalar, I think that storing this value to TreeInfo won't affect the performance.

@neerajprad This gist illustrates how numpyro and arviz are integrated. With this PR, we fully support arviz (I need to make a PR to arviz with those updates after 0.2.1 is released)! Cheers, :D

I think it is good to make a tutorial which shows how simple it is to run MCMC in numpyro, and showcase the support for arviz. WDYT?

@neerajprad
Copy link
Member

neerajprad commented Sep 30, 2019

@neerajprad This gist illustrates how numpyro and arviz are integrated. With this PR, we fully support arviz (I need to make a PR to arviz with those updates after 0.2.1 is released)! Cheers, :D

That's great! @null-a - I think we should have these utilities available within the next release for numpyro and arviz.

I think it is good to make a tutorial which shows how simple it is to run MCMC in numpyro, and showcase the support for arviz. WDYT?

Sure! We can modify the existing examples too. I think one good example is the existing one on non-centered parametrization which can be converted into a tutorial that demonstrates this integration and shows how to use diagnostic information to detect issues with the model.

@neerajprad neerajprad merged commit 72ac08f into pyro-ppl:master Sep 30, 2019
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.

Higher level interface for predictions and arviz integration
2 participants