Scalable way to train graph embeddings

Question

I have a use-case to train graph embeddings, looking for a way to do it in PyTorch and TensorFlow. The restriction is the methodology should incorporate edge weights in calculating transition probability. Can someone please advise the best framework/library to do? The node2vec implementation by the authors is in spark only, however, my current ML pipeline is more PyTorch/TF friendly.

Answer 1

Don't know if this is what you need but I know of the Ampligraph library:

Python library for Representation Learning on Knowledge Graphs