Max Balandat
Max Balandat
by passing a `150 x 3` tensor to each of the models, you're evaluating the joint posterior across all `150` points. And since this is a multi-task model with 100...
> Ps: maybe I am not understanding the concept of the 'rank' parameter but leaving the default value which I think that is zero, gives me worse results when training...
That would be great. Cc @sdaulton
It would definitely be of interest!
Constraints are scalar-valued, You'll want to specify multiple scalar objectives rather than specifying a vector-valued constraint. Something like ``` constraints=[ lambda Z: Z[..., 5], lambda Z: Z[..., 6], ] ```...
> I believe there is some bug/API issue preventing it being merged I don't think there is an API issue; basically there were some issues with the fitting stability (and...
> Would you want it like this PR as a utility function as opposed to a Class with overridden fit method? Hmm so our classes don't actually have a fit...
> I'm slowly being convinced that kernel flexibility is important. That said, I don't know how _broadly_ useful very flexible or niche kernels are for the larger community. Definitely going...
I feel like so long as the constraints are not across the elements of the q-batch ("inter-point"), but across the dimensions of each point in both the q-batch and the...
> Ax should not error when provided constraints that are not inter-point I don't think that Ax currently supports any kind of notion of inter-point constraints, only intra-point constraints. It...