Freddy Boulton
Freddy Boulton
Thanks @tcuongd ! One reason the confidence intervals with Numpyro may be wider is that I have not yet used the values in `stan_init` to initialize the sampling. I need...
@bletham Do you have any thoughts on this? Thank you!
Thank you for the thoughtful review @tcuongd ! I really appreciate it. I will address your comments shortly!
@bletham Do you have any thoughts on this? I want to know if this kind of change is mergeable before continuing to work on it. Thank you!
@bletham Have you had a chance to review this?
@jeremyliweishih Thank you for reopening ! I've renamed it to be more general and also linking with #2997 which I think is the root cause of these kinds of issues.
We should prioritize this as I think there may be some performance implications to this. Repro 1. Comment out this line in `_create_fast_fina` ```python def _create_fast_final(self): estimators = [ estimator...
The model understanding plots don't look great in dark mode either:
Now that I think about it, it's possible to do #1944 without doing this so I'm removing it from the Mar A sprint. I'm keeping it open for two reasons:...
@ParthivNaresh Can we close this PR?