All-nan value problem for the latest version

[yqdleiyi]

Hello, I found that when I run pykan model for training some dataset under version 0.2.8, all training loss would turn to nan value, adn I couldn't fix it unless I manually downgraded it to 0.2.7. Does anyone suffer from similar problem?

[x39826]

same problem, cpu is ok, but cuda will be all nan

[dark-wind]

I face same problem, and I have to just use cpu

[yxlu1222]

Hello, is there any way to fix this because the CPU is just too slow

[yxlu1222]

同样的问题，cpu 还可以，但 cuda 将全部为 nan

Is there any way to use both the CPU and the GPU to avoid errors?

[yxlu1222]

我面临同样的问题，我只能使用 cpu

Is there any way to use both the CPU and the GPU to avoid errors?

[Amursen]

i think u most watch in examples singularity part guys