-
-
Notifications
You must be signed in to change notification settings - Fork 397
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
Consider changing rcParams
default for log_likelihood
to False
#2155
Comments
Hi, I am a first time contributor! Would this be a good first issue that I can assist with? |
Is pymc still using arviz converter? |
@ilaiyengar no, sorry. This issue still needs discussion and there is no clear fix yet. You should take a look at Beginner or let us know the areas within ArviZ you are interested in on gitter so we can find something that fits your interests.
The converter to InferenceData for pymc v4 is now in pymc codebase, but for now it maintains compatibility with ArviZ, uses it's rcParams, uses dict_to_dataset... I have checked the converters and the When that happens I generally need to exit the python interpreter and end up needing to resample. Therefore, given it is an rcParam that can be set again to true on a user or project basis, I think it would make sense to set the default to false, but it will probably confuse users for a while when they try to call If set to false (or whatever we choose really) it might also make sense to split the rcParam into two in order to preserve Stan behaviour where true doesn't always compute and store pointwise log likelihood and it does only if the |
arviz/arvizrc.template
Line 10 in 2d88638
Paraphrasing from the PyMC issue: pymc-devs/pymc#6266
Would require coordinating with other PPLs that use this variable to see if they agree and can help users make the transition.
CC @OriolAbril
The text was updated successfully, but these errors were encountered: