BUG: Missing constant_data in InferenceData when defining observed data with same dims.
#7614
Labels
Comments
|
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Describe the issue:
I have defined my data in a
pm.Datacontainer along with its dimensionality(dims=)but when i use these points as likelihood and also pass the dimensionality they are no longer in theconstant_dataof theInferenceDataobject. A workaround is to avoid defining dimensionality in pm.Normal like this:obs = pm.Normal('obs', mu=mu, sigma=sigma, observed=data).I think this is a bug from newer versions of pymc because i had no problems with pymc 5.5.
I tried to debug the issue and could be in
pm.backend.arviz.find_constantswhere it checks if an observed variable could be part of the generative graph.Reproduceable code example:
Error message:
--------------------------------------------------------------------------- AttributeError Traceback (most recent call last) Cell In[8], line 17 14 obs = pm.Normal('obs', mu=mu, sigma=sigma, observed=data, dims='data_dim') 15 idata = pm.to_inference_data(prior={}) ---> 17 assert 'data' in idata.constant_data AttributeError: 'InferenceData' object has no attribute 'constant_data'PyMC version information:
Context for the issue:
No response
The text was updated successfully, but these errors were encountered: