Closed lucasxteixeira closed 1 year ago
Hi, I'm getting an error when using the evaluate/callr for encapsulation. Here is reproducible example:
library(magrittr) library(mlr3verse) learner <- mlr3::lrn("classif.ranger", predict_type = "prob", predict_sets = c("train", "test")) learner$encapsulate <- c(train = "evaluate", predict = "evaluate") learner$timeout <- 300 learner$fallback <- mlr3::lrn("classif.featureless", predict_type = "prob", predict_sets = c("train", "test")) learner <- mlr3tuning::auto_tuner( method = mlr3tuning::tnr("random_search", batch_size = 5), learner = learner, resampling = mlr3::rsmp("cv", folds = 3), measure = mlr3::msr("classif.ce"), search_space = mlr3tuningspaces::lts("classif.ranger.default"), term_evals = 50 ) gr <- mlr3pipelines::po("scale") %>>% mlr3pipelines::po("imputemean") %>>% learner cv <- mlr3::rsmp("holdout") bmr_grid <- mlr3::benchmark_grid( tasks = mlr3::tsk("breast_cancer"), learners = gr, resamplings = cv ) bmr <- mlr3::benchmark(bmr_grid)
Hey, sorry for the late reply. The timeout must be set individually for train and predict.
learner$timeout <- c(train = 300, predict = 300)
Hi, I'm getting an error when using the evaluate/callr for encapsulation. Here is reproducible example: