I have simple model like this
model input (3 32 32)
Sequential(
(conv1_A): adap_conv2d(3, 12, kernel_size=(11, 11), stride=(1, 1), padding=(5, 5), dilation=(False,False))
(conv1_A_batch_norm): BatchNorm2d(12, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(conv1_A_leaky): LeakyReLU(negative_slope=0.1, inplace=True)
)
without BatchNorm2d everything is OK , when I add BatchNorm2d I get the runtime error:
RuntimeError: dilation should be greater than zero, but got [0, 0]
I don’t use dilation anywhere why this error happen? and what that say???
thank’s for any help
The dilation of (1,1) means no dilation.
Replace (False,False) to (1,1) .
Thank you , problem solved
in my inherited class from Conv2d (named adap_conv2d) parameter disorder in constructor caused this problem, your hint solved that