Dimension mismatch in the normal likelihood with theano variables

Questions version agnostic
,
5

Hello @ricardoV94, yes, I get the error message where the likelihood is defined. I believe it’s a fairly simple problem as far as the modelling point is concerned. There’s a mean variable mu_y (1,26), there’s a sigma_y (1,26) that represents the behavior about each mean value and then there’s the observed data training_data (26,26). I want to know why setting pm.Data() is causing problems while defining the likelihood.

Here’s the result for both the variables X1 and X2:

print(X1.broadcastable)
(False, False)
print(X2.broadcastable)
(False, False)

Thanks for helping out.