Open gorlando04 opened 2 years ago
It is certainly in the category of "in development" rather than ready for mainstream use. It does also require dask, and ideally dask.distributed, including setting up a dask cluster with workers before running, which may be part of the issue.
I'm doing a research about distributed ways do build the kNNG, and i am especifically studying NNDescent. When i tested the distributed-pynndescent, that is in the fork, i couldn't perform it. Is this algorithm ready or it is on development?