Samuel Stanton

Results 15 comments of Samuel Stanton

closing the issue as the fix has been merged. Thanks again!

good suggestion, working on it

this is a rough solution ``` class DistributableStandardize(Standardize): r"""Standardize outcomes (zero mean, unit variance). This module is stateful: If in train mode, calling forward updates the module state (i.e. the...

Seems strange to me too. It's up to the BoTorch core team how you want to deal with this, it just took me a while to figure this out so...

This fix doesn't exactly replicate the earlier behavior. Previously if both the inducing points and the inputs were shared calling `VariationalStrategy.__call__(model, inputs)` would result in a 2D tensor being passed...