Added optional **kwargs to ExactMarginalLogLikelihood.forward call, so that optional keyword arguments can be passed on to the likelihood model. This is required, for example, when using it with FixedNoiseGaussianLikelihood on data that are not part of the training set, as mentioned in this issue: #2516 .
Added optional
**kwargstoExactMarginalLogLikelihood.forwardcall, so that optional keyword arguments can be passed on to the likelihood model. This is required, for example, when using it withFixedNoiseGaussianLikelihoodon data that are not part of the training set, as mentioned in this issue: #2516 .