Max Balandat
Max Balandat
Interesting. I don't run into this error (this is on pytorch and gpytorch current dev version). My first guess would be that this is either some change in type promotions...
This is most likely the `LKJCovariancePrior` over the intra-task correlation matrix, defined by default here: https://github.com/pytorch/botorch/blob/fd30429726e2067e097a0e57123cad335782de46/botorch/models/multitask.py#L476 If you trace this down this is registered here: https://github.com/cornellius-gp/gpytorch/blob/d171863c50ab16b5bfb7035e579dcbe53169e703/gpytorch/kernels/index_kernel.py#L71 Basically this would need...
cc @j-wilson who I just talked to about LKJ priors earlier today...
@sdaulton could you take a look at this one, please?
@alantan-SQ40 could you please provide a full repro so we can debug? Thanks!
Failures seem to be due to some issues with pyro on py3.10.
Closing this in favor of #1380
So we work with a lot of pretty noisy data, and so our default prior for the noise reflects that (that said it may be a little overzealous given that...
It's certainly true that the default priors may not work particularly well in some situations. Generally it won't be possible to choose default priors that work uniformly good across all...
Thanks @mc-robinson, this is a great start. Will take a closer look in the next couple of days and help out with the shape challenges. Re the test: This should...