Closed sgaist closed 1 month ago
Do we want that to be mandatory or an option ?
I get that this was always mounted in the previous implementation though I am not sure this is always needed. Any inputs ?
Do we want that to be mandatory or an option ?
I get that this was always mounted in the previous implementation though I am not sure this is always needed. Any inputs ?
Actually I think @SalimKayal is right, it is probably better to not always have this. I know people want this when they use tensorflow/pytorch but whether they always need it or only when they are doing something fancier like using multiple gpus I cannot remember.
@sgaist what do you think?
@olevski I agree as well, it will also make the session lighter on resource.
Thanks this looks good. But there is a merge conflict. So let me know and I will re-approve when you resolve it.
Done, the tests were conflicting :-)
Describe your changes
This PR implements two things:
Issue ticket number and link
Closes #630 Part of https://github.com/SwissDataScienceCenter/renku/issues/3679