-
-
Notifications
You must be signed in to change notification settings - Fork 409
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 warmup groups in from_pymc3 #1171
Conversation
Codecov Report
@@ Coverage Diff @@
## master #1171 +/- ##
=======================================
Coverage 93.11% 93.11%
=======================================
Files 94 94
Lines 9322 9339 +17
=======================================
+ Hits 8680 8696 +16
- Misses 642 643 +1
Continue to review full report at Codecov.
|
looks like pylint doesn't want us to access |
Ready for review or even merge Also related to #1083, this starts storing number of tuning steps in inference data as attrs of posterior and sample stats (both samples and warmup). |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks Oriol! LGTM but I just had two questions.
assert idata.posterior.sizes["chain"] == 2 | ||
assert idata.posterior.sizes["draw"] == 200 | ||
if save_warmup: | ||
# pylint: disable=protected-access | ||
assert idata._warmup_posterior.dims["chain"] == 2 | ||
assert idata._warmup_posterior.dims["draw"] == 100 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
- Just out of curiosity: why is the attribute to access different for
posterior
(sizes
) and_warmup_posterior
(dims
), whereas you want to see the number of chains and draws in both cases? - Also, does it mean we keep the
_
beforewarmup
?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
_ should stay before warmup. it is not a mcmc draw (in hmc; maybe) so it is only for advanced users
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Regarding first question, both attributes do the same, the test Michael sent had sizes and I used dims to make it more xarray-y but missed half of them 🤷
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@ahartikainen how opposed would you be to removing the _
before warmup? Even if only for advanced users it is still for users, the default would still be data.save_warmup = false
so unless they change it they would not have warmup groups, I feel like hiding them so much makes harder the live of advanced users without much gain for novices.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Not too opposed.
Co-authored-by: Michael Osthege <zufallsprinzip@hotmail.de>
I think it is ready to merge |
Agreed 👍
Le mer. 6 mai 2020 à 23:02, Oriol Abril <notifications@github.com> a écrit :
… I think it is ready to merge
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#1171 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AHIJMTAPLID3JKE3AEQH7X3RQHF53ANCNFSM4MY3TLBQ>
.
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Description
Support warmup groups in
from_pymc3
. It should already work, but I have not figured out the dependencies of the whole thing (it may only work on pymc3 master) nor added any test yet.Will fix #1146
Checklist