Multi-labeled edge classification in the homogeneous graph

[songsong0425]

Greetings, thank you for your effort in updating the PyG library. I have a question about the method for multi-labeled edge classification.

Unlike the basic link prediction using binary classification (0, 1), I'd like to predict the different types of edges from the homogeneous network using GAT. For example, each edge has one of the three types of relation and if it needs, it has 0 labels for the negative sample (i.e., 0, 1, 2, 3 labels for edge classification).
To my knowledge, a multi-label classification task needs one-hot encoded labels and BCE loss for training. But I only know that edge_label in Data get a scalar value for the edge label.
Although I considered using RGCNconv for this job, I'm not sure whether it is proper for this job. Is there any example for this task? Or do you have any idea? I'll appreciate your help.
Thank you for reading!

[songsong0425]

Sorry for closing this thread. I'll update this question with the code in #7873.