HaoYin
HaoYin
@collwe can you share the preprocessing programs for Pinterest
The output order of the nodes is arranged by the frequency of nodes in sampled walks, from high to low.
From the code in lines 50~51 of `src/node2vec.py` , we can get a message that all nodes or sub-graphs will be trained with this model. So, you don't need to...
Using gensim(without Spark Graphx ), node2vec can train up to 21 million nodes.