issues
search
AntonioTepsich
/
Convolutional-KANs
This project extends the idea of the innovative architecture of Kolmogorov-Arnold Networks (KAN) to the Convolutional Layers, changing the classic linear transformation of the convolution to learnable non linear activations in each pixel.
MIT License
694
stars
68
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
KAN conv instead of ResNet conv
#14
woodszp
opened
2 weeks ago
0
Dynamic allocation of Device based on the input's device to account for DDP
#13
ashimdahal
closed
1 month ago
0
KANLinear is based on ?
#12
kunliu916
opened
1 month ago
0
[feat] Is it possible to run it with non-square kernels?
#11
suzannejin
closed
1 week ago
1
Is it able to implement convolutional kan on cpu in a parrallel mode?
#10
jingyao16
opened
2 months ago
0
Questions about convolution operation
#9
zhongwlf
closed
2 months ago
2
is the b = [kernel_side // 2, kernel_side// 2] used in convolution.py at line 11 or for other usage?
#8
zhangjy2008327
opened
3 months ago
0
I tried to modify ResNet18 using CKAN, but encountered a gradient computation failure issue
#7
icadada
opened
3 months ago
0
Nit: Relative imports of KANConv script
#6
ubitquitin
opened
3 months ago
1
How to set the bias
#5
entuo
closed
3 months ago
1
How to set the number of output channels for a tensor?
#4
SkyWxxk
closed
3 months ago
1
Cifar10
#3
ernleite
closed
3 months ago
1
Great! How about the train/infer time?
#2
yuedajiong
closed
3 months ago
2
Typos in README
#1
jorisSchaller
closed
3 months ago
1