paradox
paradox copied to clipboard
mlr-org
ParamHelpers Next Generation
https://stackoverflow.com/questions/74462621/rewriting-paramset-ids-from-mlr3paradox?noredirect=1#comment131836530_74462621
### Description I'm considering replacing irace custom parameter representation with paradox. This will benefit irace since I expect your implementation to be of higher quality and it will benefit mlr3...
In the cubist regr learner, there is a parameter whose default is `sample.int(4096, size = 1)` which should be displayed in the help page of the learner constructed by `rd_info`....
```r l = lrn("classif.ranger") # the Learner class should do this: l$param_set$context_available = "task" l$param_set$values$mtry = to_tune(p_dbl(0, 1, trafo = function(x) ContextPV(function(task) max(1, round(length(task$feature_names) * x)), x) )) at =...
When a graph learner is deep cloned, the dependencies in the parameter set disappear in the original and cloned learner. ```r library("mlr3") library("mlr3pipelines") library("paradox") graph_learner = GraphLearner$new( lrn("classif.rpart")) graph_learner$param_set$add_dep("classif.rpart.cp", "classif.rpart.keep_model",...
For parameter sets with functional parameters the parameter table contains the function source code as can be seen in the example below ``` r library(mlr3extralearners) learner = lrn("classif.C50") learner$help() ```...
Either "requires" tag should be ignored for params with dependency, or dependent params with requirements should be forbidden. Currently "requires" is checked in `get_values()` and dependencies ignored.
I do not expect this PR to get merged, therefore I open it as a draft. Provides two functions `cs_to_ps` and `ps_to_cs` that allow for ConfigSpace ParamSet mappings (relying on...
'num_nodes' hyperparameter in Param_set of 'surv.deephit' learner is a ParamUty class with default value: 32, 32. To optimize that with two random num_nodes, I wrote the code below: `search_space
