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'm not very familiar with this feature of numpyro so could you say a few more words about what this is supposed to be doing and what behavior you want?
One note: I don't think this part of the model makes much sense, and might cause problems if I understand what is supposed to happen here:
The problem here is that you can't have a plate over data points in a GP model, since the points are not independent. This means that mixture models are hard to implement using GPs, because you can't do closed form marginalization over mixture memberships in the straightforward way that you can with models where the data points are conditionally independent.
Hi @dfm,
I am building a Gaussian process model that contains discrete variables. Here is a simplified version:
Running the code gives me the following error, as enumeration adds array dimensions to the kernel hyperparameters.
Is there a workaround that would allow TinyGP kernels to be compatible with Numpyro enumeration? Thank you very much!
The text was updated successfully, but these errors were encountered: