example with future and parallel CV?

[osofr]

Is it possible to provide a quick working example for parallel CV evaluation using future back-end?

Presumably, parallel CV is working? If its not implemented yet, I am happy to close this issue, thanks so much!

[jeremyrcoyle]

Aside from the other issue you opened (#16), it's working. We've been pointing people to Henrik's future documentation here: https://github.com/HenrikBengtsson/future , as parallelizing origami is now just a matter of specifying a future plan. @ck37 also requested a parallel example though, so maybe it's worth having a minimal one, just so people can get a sense of how it works before reading more in depth about Henrik's package.

@nhejazi, can you adapt the code here: https://github.com/jeremyrcoyle/origami/blob/master/tests/testthat/test_future_plan.R into an example? At some point, it might also be worth demonstrating that cv_funs themselves can create futures, and therefore we can do nested parallelization as in this example: https://github.com/HenrikBengtsson/future#nested-futures-and-evaluation-topologies This will be the mechanism by which the new origami_Superlearner parallelizes across folds and learners.

[osofr]

Great, thanks a lot! Just to clarify, any time I use future lapply in nested loops, the parallelization will be automated and balanced across several nested layers? Will look at your links, thanks!

…

On Jun 28, 2017 20:16, "Jeremy Coyle" ***@***.***> wrote: Aside from the other issue you opened (#16 <#16>), it's working. We've been pointing people to Henrik's future documentation here: https://github.com/HenrikBengtsson/future , as parallelizing origami is now just a matter of specifying a future plan. @ck37 <https://github.com/ck37> also requested a parallel example though, so maybe it's worth having a minimal one, just so people can get a sense of how it works before reading more in depth about Henrik's package. @nhejazi <https://github.com/nhejazi>, can you adapt the code here: https://github.com/jeremyrcoyle/origami/blob/master/tests/testthat/test_ future_plan.R into an example? At some point, it might also be worth demonstrating that cv_funs themselves can create futures, and therefore we can do nested parallelization as in this example: https://github.com/ HenrikBengtsson/future#nested-futures-and-evaluation-topologies This will be the mechanism by which the new origami_Superlearner parallelizes across folds and learners. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#17 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGaPTuIDZXUq6HWMKQWLc9vrLtzI8kwsks5sIxbzgaJpZM4OIhBm> .

[nhejazi]

@jeremyrcoyle - The test has been modified (with practically minimal changes) as an example in the cross_validate function here as of cf2407a. I'm going to close this issue; feel free to re-open.