How is the performence compared to traditional community detection algorithms?

[dawnranger]

Nice work to follow. However, I didn't see the comparison against traditional community detection algorithms, such as louvain, LPA, infomap etc. Any opinions about the pros and cons between the embedding&clustering pipline and traditional methods will be appreciated.

[chxgong]

when i use you algorithm to train embedding for a network, after i fit a GMM and train community embedding, the training result always has some Nan value. I can't find the reason. Can you give me some advice? thank you.