Setting features for graphs with multiple node types

Questions
#1

Suppose I have a graph with multiple node types (for example a dgl bipartite graph). How can I set the nodes features for nodes of a specific type. I can’t use the ndata function as that only works for graphs with a single node type. Is there a way I can set node features for different node types in this case? Any help would be appreciated, thanks.

#2

See the example below

import dgl
import torch

g = dgl.bipartite([(0, 0), (1, 1)], 'developer', 'develops', 'game')
g.nodes['developer'].data['h'] = torch.ones(2, 1)
g.nodes['game'].data['h'] = torch.zeros(2, 1)

print(g.nodes['developer'])
# NodeSpace(data={'h': tensor([[1.], [1.]])})
print(g.nodes['game'])
# NodeSpace(data={'h': tensor([[0.], [0.]])})
#3

@humzaiqbal Your scenario looks like a perfect fit for heterogeneous graph. Please take a look at our tutorial here: https://docs.dgl.ai/tutorials/basics/5_hetero.html