Alessandro Flati

> for a given amount of compute. What do you exactly mean by this? That KANs are slower in training/inference?

I'm actually investigating exactly this (and more) in #99 , so I do agree with you that it's something that should be tackled.

You should put into the requirements `torch==2.3.0+cu121` or whatever cuda version you need.

Actually, latest master version is bugged, without https://github.com/KindXiaoming/pykan/pull/98 @KindXiaoming

As you see, the problem stands in line `self.scale_base = torch.nn.Parameter(torch.FloatTensor(scale_base).cuda()).requires_grad_(sb_trainable)` which in a previous (bad) tentative of allowing people to use CUDA, forced the parameter to be on cuda....

You should actually change it to `cpu`, not to `mps`.

> Also, I'm unable to run any KAN model in GPU. I send to device (cuda) both the dataset and the model but keeps giving me this error: device =...

That's strange. Could you please create a reproducible gist/snippet where I can try to reproduce your case in order to further expand the PR if needed? That would very much...

I just think, as the RuntimeError describes, you do not have to cast to float through `.float()`, or maybe cast it as double

Well, first of all, looking at the ground truth it seems to me that the minimal model should be **at least** 5-depth (like `[2, *, *, *, 1]`), since there's...