Closed CaptainCuong closed 1 year ago
Here in practice we just use one layer of GFN after the graph encoders. For the final output, we only need $x$, so actually the two lines you quoted are all corresponding to Eq.5, with Eq.6 skipped since we don't need updated node embedding.
Hi Minkai Xu,
I think there is a flaw in your mathematical notation for GNN, Equation 5 and 6 in particular. I found that the notations
m_{ij}
in equation 5 and 6 are not the same. m_{ij} in equation 5 and m_{ij} in equation 6