4203fde78e0522513258a60466f1575e161b9138,gpytorch/variational/whitened_variational_strategy.py,WhitenedVariationalStrategy,initialize_variational_dist,#WhitenedVariationalStrategy#,55

Before Change

        return kl_divergence

    def initialize_variational_dist(self):
        if not self.variational_params_initialized.item():
            prior_dist = self.prior_distribution
            inv_prior_dist = torch.distributions.MultivariateNormal(
                prior_dist.mean,
                prior_dist.lazy_covariance_matrix.add_jitter()
                .evaluate()
                .double()
                .inverse()
                .type_as(prior_dist.covariance_matrix),
            )
            self.variational_distribution.initialize_variational_distribution(inv_prior_dist)
            self.variational_params_initialized.fill_(1)

    def forward(self, x):
        
        The :func:`~gpytorch.variational.VariationalStrategy.forward` method determines how to marginalize out the
        inducing point function values. Specifically, forward defines how to transform a variational distribution

After Change

        return kl_divergence

    def initialize_variational_dist(self):
        prior_dist = self.prior_distribution
        inv_prior_dist = torch.distributions.MultivariateNormal(
            prior_dist.mean,
            prior_dist.lazy_covariance_matrix.add_jitter()
            .evaluate()
            .double()
            .inverse()
            .type_as(prior_dist.covariance_matrix),
        )
        self.variational_distribution.initialize_variational_distribution(inv_prior_dist)

    def forward(self, x):
        
        The :func:`~gpytorch.variational.VariationalStrategy.forward` method determines how to marginalize out the

Instances