onnx vs weight norm

[Selectorrr]

Hi. @Plachtaa Thank you so much for sharing your work.
I'm trying to export this model to ONNX, but I ran into a problem that Wavenet uses weight norm in its calculations, which increases the graph complexity and model weight depending on the number of diffusion steps. Is there any way around this issue?

[Selectorrr]

looks like folding weights from weight norm to Conv1d might help, but it will take time, I will check

[littlebeanhp]

looks like folding weights from weight norm to Conv1d might help, but it will take time, I will check

Have you succeeded exporting this model ? I have the same question :)

[deguodedongxi]

Any updates on this topic? I am also trying to convert the model to onnx but I am also struggling with the wavenet implementation