Fix setting initvals on ZeroSumTransform rv #7773
Conversation
|
Is a new test required for this? I can easily add the one in the issue, but it seems a bit of an overkill, considering it will sample a model. |
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #7773 +/- ##
=======================================
Coverage 92.83% 92.84%
=======================================
Files 107 107
Lines 18354 18378 +24
=======================================
+ Hits 17039 17063 +24
Misses 1315 1315
🚀 New features to boost your workflow:
|
Fixed the one below and added a few basic tests. Turns out, this transform was only tested through ZeroSumNormal before, so had to add a new segment for it in transforms |
| @@ -312,7 +312,7 @@ def extend_axis(array, axis): | |||
| def extend_axis_rev(array, axis): | |||
| normalized_axis = normalize_axis_tuple(axis, array.ndim)[0] | |||
|
|
|||
| n = array.shape[normalized_axis].astype("floatX") | |||
| n = pt.cast(array.shape[normalized_axis], "floatX") | |||
There was a problem hiding this comment.
Instead you can call pt.as_tensor(array)
There was a problem hiding this comment.
I'm wondering, is it better than pt.cast for some reason?
There was a problem hiding this comment.
Not really, I was just thinking out loud
|
One test is failing now due to precision issues? |
Description
Fix setting initvals on ZeroSumTransform'ed rv
Related Issue
Checklist
Type of change
📚 Documentation preview 📚: https://pymc--7773.org.readthedocs.build/en/7773/