node2vec is not deterministic, i.e. multiple runs will lead to different embeddings

[antonkratz]

node2vec is not deterministic, i.e. multiple runs will lead to different embeddings.

I am working with to very similar, but not identical graphs. If I compare their embeddings, it is not clear which changes are caused by the graphs being different, or simply by having done two different runs of embeddings.

Can this be somehow adressed?

Would setting the random seed to a fixed value address this?

If yes, can the random seed be set to a fixed value?