First of all, I would like to thank the TVM developers for the implementation of 3d operations such as 3D convolutions. This feature is very much appreciated by 3D CNN users like me.
I am contacting you because of the lack of support for group convolutions. For example, if you try to import a network with a group convolution, you get the following error :
No support for arbitrary group number for conv3d
which is cause by groups != 1 in https://github.com/apache/incubator-tvm/blob/28ea54aa9d09f5087ef9ee467168a3c6c596a336/python/tvm/relay/op/strategy/generic.py#L470
Is there a way to solve this problem? Do you have any plans to include this feature in the TVM development roadmap ?