tfkanSome doubts about tfkan

[Summer-77]

Hello, I would like to ask you about tfkan in which I input my training data and it goes to fit it, but when I go to test it after the training is over, does the fitted function still exist? As we all know that tfkan is a network similar to MLP, do we need to train it before using it and then embedding it in our own model. I hope you can answer this doubt for me.

[ZPZhou-lab]

Yes, the fitted function still exist after you train the model. All of the tfkan modules have the same behavior as modules in Tensorflow. Therefore, you can build your model using tfkan and other tensorflow modules, train it and use it at testing time.

[Summer-77]

I've heard that kan is a bit friendlier when it comes to timing signals, but it doesn't seem to work very well when I add it, is it because I have a problem with the parameter settings?