# Can GraphRNN be used with very large graphs?

In the GraphRNN paper, the authors only implement the algorithm up to a graph size of 2k nodes. Would this still work on much larger graphs (on the order of $$10^7$$)? Or would the computation just become too substantial?