-
-
Notifications
You must be signed in to change notification settings - Fork 2.1k
Make VI work on v4 #4582
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
Make VI work on v4 #4582
Conversation
@ferrine I understand this is still WIP ? After #4579 (the rebase) replaces |
93a096d
to
c9fa127
Compare
@ferrine this was one of the last remaining PRs that still had |
I included the VI tests as a standalone job into the pytest workflow, so it's output is not contaminated by flaky tests that still exist on Running locally on Windows:
List of test failures
Hope this helps :) |
@brandonwillard I have a subset of functionality working. AEVB inference does not work and I start thinking maybe I mark this as NotImplemented?
|
I agree that we can skip that.
…On Wed, Jul 28, 2021 at 4:47 PM Maxim Kochurov ***@***.***> wrote:
@brandonwillard <https://github.com/brandonwillard> I have a subset of
functionality working. AEVB inference does not work and I start thinking
maybe I mark this as NotImplemented?
if aevb: raise NotImplementedError("AEVB is not yet supported, open an issue once you need it https://github.com/pymc-devs/pymc3/issues")
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#4582 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAFETGDXT7YGZKQLTFRWAITT2AJ75ANCNFSM4Z7ATPNQ>
.
|
Hmm, the issue still appears with grouped inference that might be some shape bug/typo somewhere. I'll keep working on this. This should not be that complicated I hope |
Codecov Report
@@ Coverage Diff @@
## main #4582 +/- ##
==========================================
+ Coverage 81.44% 87.21% +5.76%
==========================================
Files 81 81
Lines 14204 14265 +61
==========================================
+ Hits 11569 12441 +872
+ Misses 2635 1824 -811
|
/pre-commit-run |
@ferrine I rebased this. |
One weird test is left and it's done. Also I remember to change skip to xfail for one test |
Tests are passing 🥳! You need to rebase from the rename, however. |
Co-authored-by: Ricardo Vieira <28983449+ricardoV94@users.noreply.github.com>
Co-authored-by: Ricardo Vieira <28983449+ricardoV94@users.noreply.github.com>
Co-authored-by: Ricardo Vieira <28983449+ricardoV94@users.noreply.github.com>
@ferrine I pushed some changes and closed outdated conversations. Can you go through the remaining open conversations and address/close them? Otherwise I think this is pretty ready to merge, and we can focus on refactoring/deprecating the remaining methods in follow up PRs |
@ricardoV94 are we good to merge? |
Only one way to find out... |
I commented one instance where detailed type info from the docstring could be annotated in the signature. |
Co-authored-by: Michael Osthege <michael.osthege@outlook.com>
/pre-commit run |
/pre-commit-run |
No it's something that should be fixed on main, I merged a PR too wrongly. |
Incredible -- congrats @ferrine and @ricardoV94! |
VI in V4
Referencing #4523 Here to work on VI module
Step 1
The first goal of this PR is to fix test errors in
because there are a lot of errors
Step 2
TBD