Wesley Maddox

At this point, you can use `LMCVariationalStrategy` to model correlated outputs -- which I don't believe existed back when this issue was written. A tutorial on this is in the...

Not quite, this corresponds to a linear model of co-regionalization (LMC) rather than an intrinsic model of co-regionalization (ICM) (which is what the MultitaskKernel implements). This implementation reduces to an...

This is because `SoftmaxLikelihood` actually doesn't expect one-hot encodings but rather class numbers. Changing the targets to `y_train = torch.where(y_train)[1]` resolves the error.

Not really, although it looks like from the [documentation of sklearn](https://scikit-learn.org/stable/modules/generated/sklearn.gaussian_process.kernels.ConstantKernel.html) that you can use it as either an outputscale for the kernel analogous to gpytorch's `ScaleKernel` or as an...

I don't think I follow what you're trying to do here. Why couldn't you pass in a callable that just explicitly computes the posterior on task 1? something like: ```python...

I believe that there's utilities for returning only one task during the posterior call in Botorch's `MultitaskGP` models. I think I get a sense about what you're trying to do...

For one, I don't think you'd need to implement anything custom here: you could just define the kernel as a function of a standard kernel: ```python self._covariance_module = RBFKernel() def...

My issue is that in general, i'm pretty sure K(x - y) isn't positive definite any more. see the counter example https://github.com/cornellius-gp/gpytorch/discussions/1894 . Can you evaluate the covariance matrix when...

Basically yes, although it depends on if the multi-task MVN is `_interleaved` or not. See [here](https://docs.gpytorch.ai/en/stable/distributions.html#multitaskmultivariatenormal) for an explanation. But basically, if the covariance is interleaved then for each data...

Can you provide a model class here? From what I see, the tutorial doesn't include the additive kernel portion so it'd be nice to see what exactly is going wrong.