Max Balandat
Max Balandat
> is there a reason I cannot pass a covar_module parameter to HeteroskedasticSingleTaskGP? I guess ideally it seems I would want to be able to control the kernel of both...
One thing to be aware of is that the models derived from `BatchedMultiOutputGPyTorchModel` have some magic going on that automatically moves a multi-output dimension into a separate model batch dimension....
@mc-robinson This is awesome, many thanks for setting this up and doing the benchmarking. I'd love to see this in a PR. Since this really is a training procedure that...
@mc-robinson, have you made any progress on this PR? I think this is pretty nifty and I'd like to get this in soon. We can also package this up ourselves...
If you can put up an PR early next week that would be great. It doesn't have to be complete, we can take care of cleaning it up. Good luck...
Yeah. I have an internal set of changes with an updated version but it has some (surmountable) issues and needs to be rebased.
cc @izmailovpavel
Yeah, I think it would be great to have that. I do concur with @saitcakmak on this not being a straightforward extension of what we do for the input transforms.
Thanks for flagging, this is a great observation. It is indeed a problem that "small step" means different things in different directions, we really want this to be isotropic. I'm...
> Additionally, is there a way to visualise or know that the following candidates predicted are correct before trying out the experiment? What do you mean by "correct"? You can...