Could you tell me how the code corresponds to the paper please?

[LoyalBlanc]

For example, where is the code "Adaptive graph scattering decomposition" and "Adaptive graph spectrum aggregation"?

The code like ` N, J, V, d = y.shape

    y_conv = self.bn1_conv(y[:,0].contiguous().view(N, -1)).view(N, 1, V, d)
    y_scat = self.bn1_scat(y[:,1:].contiguous().view(N, self.J * self.J, -1).transpose(1,2)).transpose(1,2).view(N, self.J * self.J, V, d)`

is far beyond my imagination since it view a tensor with (N, J, V, d) to (N, J*J, V, d) ...?

[catening]

For example, where is the code "Adaptive graph scattering decomposition" and "Adaptive graph spectrum aggregation"?

The code like ` N, J, V, d = y.shape

    y_conv = self.bn1_conv(y[:,0].contiguous().view(N, -1)).view(N, 1, V, d)
    y_scat = self.bn1_scat(y[:,1:].contiguous().view(N, self.J * self.J, -1).transpose(1,2)).transpose(1,2).view(N, self.J * self.J, V, d)`

is far beyond my imagination since it view a tensor with (N, J, V, d) to (N, J*J, V, d) ...?

哥们你好哇，这代码CMU数据集和3DPW数据集跑出来效果和他一样嘛

[ihavenotgoodname]

can you please tell me how to test the model?I can't find it in opt file