KindXiaoming/pykan

Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cuda:1!

Opened this issue · 0 comments

alikeayal commented

Use MultKAN.py and raise this Error.

I check the debug, The variable x is on CUDA: 1, while variables such as subnode_stcale are on CUDA: 0.

daccea9edfc9d9f2c52c87b8f8adb62
77729d11acc4ced48d5ae1533c9b99e
aeed1cb1f993324c4ac13cb85db55c8

I just updated pykan today, starting from version 0.0.5.
I was planning to try using KAN on multiple GPUs, but unexpectedly encountered this issue.
I hope someone can help me solve this problem. Thank you！ @KindXiaoming