Clarification on Distributed Dask ML (Is ML really Distributed?)

Hello all, going thought the official docs.
And it seems that Dask can only run ML one 1 machine in parallel?

Are we able to setup Scikit Learn and compute across worker nodes connected to a central scheduler ?

I know it can leverage a local cluster.


Hi @gavan1, welcome!

Sure, if you’ve got a client connected to a Dask Cluster, then Scikit Learn with Dask backend will automatically use this cluster for executing jobs!

What makes you think in the docs that this is not possible?

1 Like