pykan

pykan copied to clipboard

Ikeda Map Learning

3

First off, great work! Second, I'm trying to learn the "Ikeda" map using KAN. My question is do you have any suggestions on improving this? Also this runs too slow...

AminMoradiXL

Strange behaviour of very trivial toy example

6

Here's the gist showing a commented jupyter notebook demonstrating how KAN is failing at a very simple task, presumably because involving large numbers as outputs. https://gist.github.com/AlessandroFlati/a5e6c5868bc4d67d51e99a7a5ec63322

AlessandroFlati

np.sin(3*i)*np.random.binomial(n=1,p=0.5,size=1) binomial其他也行 封闭曲线 多值函数 带有阶次和求和的，比如Weierstrass等这种形式的 ...... 我有一个问题： 对于复杂问题，选择基函数困难，还是函数复合形式困难？ 这个问题背后的一个例子就是： 采用傅里叶级数的做法，可以用若干不同周期组合起来，画出非常漂亮好玩的图画。 这里基函数也不用选了，单反过来，假设给定任意一个人的一张脸，要求出这个函数复合形式，感觉就很难的样子。 （我没有深入研究过这个，知识凭直觉。）

yuedajiong

Grid learning with gradient descent

1

My question is about this line in the paper: "Other possibilities are: (a) the grid is learnable with gradient descent, e.g., [16];" There is an implementation of this? Also, I...

Indoxer

as title, thanks. including: scale_spline, c: c = coef2curve(coef=self.coef[self.weight_sharing], x_eval=i, grid=self._grid[self.weight_sharing], k=self.k).permute(1,0) self.scale_spline.unsqueeze(dim=0) * c

yuedajiong

How to Extract Predicted Values

2

Firstly, I would like to congratulate the KAN team on their outstanding work. I am currently implementing the KAN for genetic data. My question is, how can we extract the...

rxa615

This will solve CPU-only, CUDA-only and any mix of them.

3

This solves the post-`fix_symbolic` problem with cuda, the `initialize_from_another_model` problem with cuda, and the `cpu` problem related (already mentioned in [this PR](https://github.com/KindXiaoming/pykan/pull/91)) that forced to use cuda.

AlessandroFlati

Corresponding explanation

6

Hello, what is the proof that kan and mlp are dual?

ghost

Alternative Function Approximation Methods? In Re: Performance

3

It seems that the current 'issue' with KANs is that they appear to be less performant in training for a given amount of compute. Perhaps there are alternative ways to...

zjcurtis

M1 runtime fails with "AssertionError: Torch not compiled with CUDA enabled"

14

Hi! Thanks a lot for the awesome paper and implementation! I can't get it to run on my M1 machine. I built `pytorch` from source, with disabled CUDA options, as...

rmrfxyz