15b3563c3c7f700ee0bf23b0477527dfa5323cea,test/examples/test_spectral_mixture_gp_regression.py,SpectralMixtureGPModel,__init__,#SpectralMixtureGPModel#Any#Any#Any#,29

Before Change

    def __init__(self, train_x, train_y, likelihood):
        super(SpectralMixtureGPModel, self).__init__(train_x, train_y, likelihood)
        self.mean_module = ConstantMean(constant_bounds=(-1, 1))
        self.covar_module = SpectralMixtureKernel(
            n_mixtures=3,
            log_mixture_weight_bounds=(-5, 5),
            log_mixture_mean_bounds=(-5, 5),
            log_mixture_scale_bounds=(-5, 5),
        )

    def forward(self, x):
        mean_x = self.mean_module(x)
        covar_x = self.covar_module(x)

After Change

    def __init__(self, train_x, train_y, likelihood):
        super(SpectralMixtureGPModel, self).__init__(train_x, train_y, likelihood)
        self.mean_module = ConstantMean(prior=SmoothedBoxPrior(-1, 1))
        self.covar_module = SpectralMixtureKernel(
            n_mixtures=3,
            log_mixture_weight_prior=SmoothedBoxPrior(
                torch.ones(3).fill_(-5).exp_(), torch.ones(3).fill_(-5).exp_(), sigma=0.1, log_transform=True
            ),
            log_mixture_mean_prior=SmoothedBoxPrior(
                torch.ones(3).fill_(-5).exp_(), torch.ones(3).fill_(-5).exp_(), sigma=0.1, log_transform=True
            ),
            log_mixture_scale_prior=SmoothedBoxPrior(
                torch.ones(3).fill_(-5).exp_(), torch.ones(3).fill_(-5).exp_(), sigma=0.1, log_transform=True
            ),
        )

    def forward(self, x):
        mean_x = self.mean_module(x)
        covar_x = self.covar_module(x)

Instances