[cluster] Optimize jh-cuda-python image for size

src-d / ml-backlog

Issues belonging to source{d}'s Machine Learning team which cannot be related to a specific repository.

0 stars 3 forks source link

Closed vmarkovtsev closed 5 years ago

vmarkovtsev commented 5 years ago

Out k8s setup loads the whole image into memory. It is currently 8 gigs. We need to explore how the size can be reduced.

vmarkovtsev commented 5 years ago

I did what I could.

vmarkovtsev commented 5 years ago