Jupyter notebook kernel dies while computing ggvec embeddings

[vhranger]

Jupyter notebook kernel dies while computing ggvec embeddings

[VHRanger]

Hi, Sorry I'll only have time to fix this next week.

0. Try manually making the NX graph into a CSRGraph by making G = cg.csrgraph(G). See if it crashes there.

1. Try running the code in a normal .py file and monitor the RAM usage

2. If the RAM usage is normal try running GGVec for one epoch with the smallest possible settings (embedding dimensions 1 or 2, negative_ratio = 0, etc.)

Please report back what you find

[janvekarnaveed]

So the very first step of making a networkx graph object itself fails

[VHRanger]

RIght, you're running out of RAM making the networkX object then.

This is because networkX takes a large amount of memory per edge and per node, so trying to make a 7m edge graph on a machine without >1TB of RAM will crash the computer from lack of memory.

Since this particular issue is with networkX I'll close it (The other two are on my side).