This message was deleted.

Hello! Are you looking for decomposition of the models? Maybe this tutorial can help: https://nixtlaverse.nixtla.io/neuralforecast/examples/signal_decomposition.html Basically, N-BEATS has a stack that is responsible for trend, and another for seasonality, and there are ways to plot those in neuralforecast. Same thing with N-HiTS where each stack specializes in different frequencies of the data. I'm not sure if it answers your question, but I hope it helps!

Thank you, but actually I think my question is a little different. With tensorflow I do it this way:

from tensorflow.keras.utils import plot_model
plot_model(lstm, to_file='model_plot.png', show_shapes=True, show_layer_names=True)

but with pytorch and the Nixtla model I couldn't do it. The result of this code is this print:

To my knowledge, you have to use another library for that, like torchviz. I haven't done it before, but here's a quick tutorial on how to use torchviz: https://www.appsilon.com/post/visualize-pytorch-neural-networks I hope this helps!

Ok, thank you!