[DOC] Better document n_jobs param in forest models #17480
Conversation
| trees. This works by creating n_jobs sets of trees and computing them | ||
| in parallel. |
There was a problem hiding this comment.
that's a little bit ambiguous IMHO, also I'm not sure that's what happens in practice. Did you want to refer to the type of scheduling that is applied, e.g. dynamic vs static scheduling?
There was a problem hiding this comment.
Thanks, @NicolasHug ! Can definitely clarify.
I hadn't thought to refer to static v dynamic scheduling -- we looked at the some of the documentation that was marked "completed" in #14228, and it didn't seem to include that level of detail. Do you have a recommendation for where to look at an example ?
There was a problem hiding this comment.
unfortunately the list in #14228 (comment) doesn't seem up to date
and in fact the docstring for the random forest was already updated in #14628
If you think the current version isn't clear enough we can definitly improve it though. What did you find confusing / unclear?
There was a problem hiding this comment.
Thanks, @NicolasHug ! I also thought the documentation was mostly quite clear -- our addition was just meant to add a bit more detail about how tree models themselves are parallelized, since the current description focused on the class methods.
But that might be the more sensible description ! Happy to close this if you think it's confusing.
There was a problem hiding this comment.
Got it, yeah in this case I think this could be closed. Sorry for the extra work!
|
@emdupre |
Reference Issues/PRs
What does this implement/fix? Explain your changes.
This PR better documents the
n_jobsparameter for forest models.Additional comments
Submitted with @annejeevan for the #DataUmbrella sprint.