How to get node embeddings from a trained .pkl?

I have successfully trained a model, but I don't know how to use the generated .pkl file. 

I want to use the node embeddings to do non-overlapping community detection.