Franck Charras
Franck Charras
> I would love this. But I believe that last time I looked, I found that it > was actually fairly technical. What would be the issues with @drhyrum code...
Thank you I'd very much like to take a look at it, will update if any improvements. Currently I have in mind to try to feed Parallel a generator that...
Thank you for opening up this thread :+1: Callbacks have become my favorite way for scripting and automating almost all what I want happening during a fit (initializations, lr management,...
I started to work on the callback task today, it breaks a certain number of tests (some because of the mock system, and by design sklearn breaks when it tries...
@tomMoral this can be closed after https://github.com/joblib/joblib/pull/1393/
This is an old thread :grin: . When I started it I had little understanding of what was really going on. It turns out naively trying to have a cache...
For reference, this is what benchopt info currently prints for me for a project with 1 dataset 1 objective and and 1 solver: ``` ---------- # DATASETS Simulated ---------- #...
Wasn't this issue related to https://github.com/joblib/joblib/issues/1473 ? if so it has been adressed in the main branch already and will be fixed in an upcoming patch release.
Thank you for the feedback, I think this can be fix quickly by adding `super().__init__` indeed. But I'm not sure why `Parallel` inherits from `Logger` here, since the methods it...
The issue should be solved after yesterday patch release. Let's keep open for discussion about possible deprecation ?