Valid or not? if manually insert 'num_classes' as a new dimension for lables

Hi, suppose my input_channel is 1, so my input looks like: [1,1,3,3,3] -->[batchSize,in_channel, dim1,dim2,dmi3]

and my num_classes is 20. So my label should be [1,20,3,3,3] in order to avoid any mismatch problems.

But I’m curious, is it valid to do that?

PS: my inputs are MRI data and have the size [dim1,dim2,dim3]. I insert the in_channel dimension myself. I also want to know if that’s valid or not.

I’m a newbie and sorry for the dumb questions. Thanks in advance!!