You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think pm.set_data jumps out as the odd one out that doesn't work with a functional approach but mutates the model inplace. Would be nice to allow pm.set_data with the same API (could introspect the first argument?). Under the hood we can use pm.do which already allows the simple behavior of pm.set_data.
We can also create some function with another name. In either case, we should not break the existing behavior of set_data.
It also seems like some of these model transforms should accept new coords to override the old ones when we want to change the shape of the variables?
The text was updated successfully, but these errors were encountered:
Description
I tried to demonstrate the basic functionality of PyMC in this notebook: https://colab.research.google.com/drive/1JyIeu2NMl_z6Y8ZniZn3jDrzzUL3i81q?usp=sharing
I think
pm.set_data
jumps out as the odd one out that doesn't work with a functional approach but mutates the model inplace. Would be nice to allowpm.set_data
with the same API (could introspect the first argument?). Under the hood we can usepm.do
which already allows the simple behavior ofpm.set_data
.We can also create some function with another name. In either case, we should not break the existing behavior of
set_data
.It also seems like some of these model transforms should accept new coords to override the old ones when we want to change the shape of the variables?
The text was updated successfully, but these errors were encountered: