A post in "What is feat_dict and how is it accessed?" requires staff attention

[AlexMRuch]

What is feat_dict and how is it accessed?

Forgive me for my ignorance, but I can’t seem to figure out what feat_dict is or how to access it in the example on https://doc.dgl.ai/tutorials/basics/5_hetero.html: def forward(self, G, feat_dict): # The input is a dictionary of node features for each type funcs = {} for srctype, etype, dsttype in G.canonical_etypes: # Compute W_r * h Wh = self.weight[etype](feat_dict[srctype]) # Save it in graph for message passing G…

I posted this question because I thought I was hitting a technical error; however, I realized that my directed network should be made undirected to work. Feel free to delete this, as I answered my own question, or (if you think it will help others) let me know if I should update the title to be about the “h” in node data and not about feat_dict. Thanks so much!

[VoVAllen]

Thanks for letting us know. We agree there is an issue and we’re looking into it.

[AlexMRuch]

Oh, no problem! Thanks for developing dgl! Absolutely love it and look forward to using it more.

Is the issue I raised in my thread really a technical issue or just a bout of confusion on my end (e.g., message passing is directed, so if no canonical edges exist that pass messages to one node type it wouldn’t have h)? I suppose that detail about directedness could be mentioned in the tutorials, but the cause of my problem was probably just coding too late at night and forgetting that detail and how message passing actually happens

Thank you for getting back to me! Please let me know if the title of the thread should be changed! Not sure if I’m able to do that or if you all need to do it.

Best,
Alex