Avoid saving big optimization path object
Opened this issue · 0 comments
PhilippPro commented
When using mlrHyperopt very big objects get saved. This can be very annoying when doing repeated cross validation on several datasets, cause big objects are saved/used.
Example:
task = makeClassifTask(data = iris, target = "Species")
lrn = makeLearner("classif.svm")
lrn = makeHyperoptWrapper(lrn)
mod = train(lrn, task)
object.size(mod$learner.model$opt.result)
Is it possible to avoid this somehow?