99f6a9b9b366e20ebc300fc5be904308c17c484f,src/spn/experiments/RandomSPNs_layerwise/rat_spn.py,RatSpn,forward,#RatSpn#Any#,284

Before Change

    def forward(self, x: torch.Tensor):
        Computes the class conditional distributions P(X | C) for each class.
        // Go over all regions
        split_results = []
        for spn in self.region_spns:
            split_results.append(spn(x).squeeze(1))

        x = torch.stack(split_results, dim=1)

        // Merge results from the different SPN into the channel dimension
        x = x.view(x.shape[0], 1, -1)

After Change

        x = self._forward_layers(x)

        // Merge results from the different repetitions into the channel dimension
        n, d, c, r = x.size()
        assert d == 1  // number of features should be 1 at this point
        x = x.view(n, d, c * r, 1)

        // Apply C sum node outputs

Instances