tomotwin-cryoet
tomotwin-cryoet copied to clipboard
Use distributed umap from rapids
This would allow multi gpu estimation of umaps.
https://docs.rapids.ai/api/cuml/stable/api/#cuml.dask.manifold.UMAP