I am apply convolution to a 4D tensor but my tensor is only 6 elements deep in the 1st dimension. After applying the first two convolution layers, I want to reshape the sparse tensor to a 3D tensor and then use 3-D convolution on that tensor. I cannot find anything on the documentation that would allow me to easily to do this.
I am apply convolution to a 4D tensor but my tensor is only 6 elements deep in the 1st dimension. After applying the first two convolution layers, I want to reshape the sparse tensor to a 3D tensor and then use 3-D convolution on that tensor. I cannot find anything on the documentation that would allow me to easily to do this.