The forward
function of GraphConv
class (here is the link), there are the codes
graph.srcdata['h'] = feat_src
graph.update_all(aggregate_fn, fn.sum(msg='m', out='h'))
rst = graph.dstdata['h']
if weight is not None:
rst = th.matmul(rst, weight)
but in the tutorial of graph attention network, the input for forward
function of GATLayer
is stored g.ndata
.
How to use ndata
property and srcdata
(dstdata
) property ?
For example, is it okay to use ndata
instead of srcdata
and dstdata
in the `GraphConv’ forward function mentioned above ?