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
We shouldn't raise errors when our measurable rewrites don't apply, because another rewrite (including a user custom one) could implement a transformation for a given graph.
However we could add debug logs for the cases where we almost wanted to raise an error because the graph is likely not measurable (e.g., if the graph looks like a convolution instead of a conditional transformation).
This way users (and devs) can figure out why PyMC refused to figure out the logprob for their graph, without having to raise eager errors or warnings.
The text was updated successfully, but these errors were encountered:
We shouldn't raise errors when our measurable rewrites don't apply, because another rewrite (including a user custom one) could implement a transformation for a given graph.
However we could add debug logs for the cases where we almost wanted to raise an error because the graph is likely not measurable (e.g., if the graph looks like a convolution instead of a conditional transformation).
This way users (and devs) can figure out why PyMC refused to figure out the logprob for their graph, without having to raise eager errors or warnings.
The text was updated successfully, but these errors were encountered: