Sebastian Fischer
Sebastian Fischer
Couldn't we also modify the `find_suggestions()` function?
Yeah, it's not done yet. Got sidetracked but will probably address this monday.
Thanks for looking into this! But tensors have reference semantics so I think calling `model$to(device)` afterwards should actually be okay (unless there is a misunderstanding from my side).
What I mean with the issue is that one could instead using something like `with_torch_local_seed(...)` to make the subsequent tests not deterministic.
@dfalbel let me know what you think. This should not be too hard to implement I guess.
> The approach sounds reasonable, my questions are: > > * Can a CRAN package take a dependency on a Bioc package? Yes, I am pretty sure. > * Can...
Also, can you maybe explain again (if you know) why this is not a problem for the pytorch dataloader?
Maybe this is also something that interests @shikokuchuo, as it might be a nice feature for mirai as well?
does this help: https://mlr3torch.mlr-org.com/articles/lazy_tensor.html? Then you just need to replace the torch_ingress_num with a torch_ingress_ltnsr
i am on vacation right gut now, so won’t be able to address these suggestions for improvements immediately, but thanks a lot!