What’s the most efficient way for calculating the rank of a weight matrix from a network?
torch.matrix_rank should work for a 2D tensor.
torch.matrix_rank