block_size for pruning 3D weights

System information
Ubuntu 18.02

• TensorFlow version (you are using):
  2.4.1
• Are you willing to contribute it (Yes/No):
  Yes
  This relates to the pruning optimization

"block_size: The dimensions (height, weight) for the block sparse pattern
in rank-2 weight tensors."

Currently it seems that the block_size feature during pruning only works for 2D weights and this seems to limit the application to dense layers. Would it be possible to block the 3D weights typically used by conv layers ? Not sure if there is any fundamental technical reason that prohibits this to work ? Since most layers are of the conv type it would speed up execution if the blocks could be used with 3D weights.

Thanks.