mlr3automl: Automated Machine Learning for 'mlr3'
Description
Flexible automated machine learning (AutoML) system for the 'mlr3' ecosystem. Automatically selects a suitable machine learning algorithm and tunes its hyperparameters for a given task. Constructs preprocessing pipelines with multiple parallel branches using 'mlr3pipelines' and jointly optimizes them together with the learners using 'mlr3tuning'. The optimization is driven by asynchronous decentralized Bayesian optimization by Egele et al. (2023) doi:10.1109/e-Science58273.2023.10254839.
Author(s)
Maintainer: Marc Becker [email protected] (ORCID) [copyright holder]
Authors:
-
Marc Becker [email protected] (ORCID) [copyright holder]
-
Damir Pulatov [email protected]
-
Baisu Zhou [email protected]
-
Lona Koers [email protected]
See Also
Useful links:
-
Report bugs at https://github.com/mlr-org/mlr3automl/issues
Auto Class
Description
This class is the base class for all autos.
Value
Object of class R6::R6Class and Auto.
Public fields
id-
(
character(1)). properties-
(
character()). task_types-
(
character()). packages-
(
character()). devices-
(
character()). n_cpu-
(
integer(1))
Number of CPUs a single training of the learner uses. n_gpu-
(
integer(1))
Number of GPUs a single training of the learner uses.
Methods
Public methods
Auto$new()
Creates a new instance of this R6 class.
Usage
Auto$new( id, properties = character(0), task_types = character(0), packages = character(0), devices = character(0), n_cpu = 1L, n_gpu = 0L )
Arguments
id-
(
character(1)). properties-
(
character()). task_types-
(
character()). packages-
(
character()). devices-
(
character()). n_cpu-
(
integer(1)). n_gpu-
(
integer(1)).
Auto$check()
Check if the auto is compatible with the task.
Usage
Auto$check(task, memory_limit = Inf, large_data_set = FALSE, devices)
Arguments
task-
(mlr3::Task).
memory_limit-
(
integer(1)). large_data_set-
(
logical(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
Auto$graph()
Create the graph for the auto.
Usage
Auto$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
Auto$early_stopping_rounds()
Estimate the number of early stopping rounds (the patience) for a learner.
budget is the maximum number of training rounds (boosting iterations or epochs) the learner may use.
The patience is capped well below the budget, otherwise early stopping and validation-based internal tuning
can never trigger and the learner always trains for the full budget.
Usage
Auto$early_stopping_rounds(task, budget = Inf)
Arguments
task-
(mlr3::Task).
budget-
(
integer(1))
Maximum number of training rounds (boosting iterations or epochs) the learner may use.
Auto$estimate_memory()
Estimate the memory for the auto.
The estimate is the host memory in MB, so learners that allocate on the gpu return -Inf.
Usage
Auto$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
Auto$finalize_model()
Prepare the graph learner for the final model fit. Called after tuning to undo tuning-only setup (e.g., timeout callbacks).
Usage
Auto$finalize_model(graph_learner)
Arguments
graph_learner
Auto$design_default()
Default hyperparameters for the learner.
Usage
Auto$design_default(task)
Arguments
task-
(mlr3::Task).
Auto$design_set()
Get the initial hyperparameter set for the learner.
Usage
Auto$design_set(task, measure, size)
Arguments
task-
(mlr3::Task).
measuresize-
(
integer(1)).
Auto$search_space()
Get the search space for the learner.
Usage
Auto$search_space(task)
Arguments
task-
(mlr3::Task).
Auto$clone()
The objects of this class are cloneable with this method.
Usage
Auto$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Catboost Auto
Description
Catboost auto.
Value
Object of class R6::R6Class and AutoCatboost.
Super class
Auto -> AutoCatboost
Methods
Public methods
Inherited methods
AutoCatboost$new()
Creates a new instance of this R6 class.
Usage
AutoCatboost$new(id = "catboost")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoCatboost$graph()
Create the graph for the auto.
Usage
AutoCatboost$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoCatboost$estimate_memory()
Estimate the memory for the auto.
Usage
AutoCatboost$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoCatboost$internal_measure()
Get the internal measure for the auto.
Usage
AutoCatboost$internal_measure(measure, task)
Arguments
measuretask-
(mlr3::Task).
AutoCatboost$clone()
The objects of this class are cloneable with this method.
Usage
AutoCatboost$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("catboost")auto("catboost")
Extra Trees Auto
Description
Extra Trees auto.
Value
Object of class R6::R6Class and AutoExtraTrees.
Super class
Auto -> AutoExtraTrees
Methods
Public methods
Inherited methods
AutoExtraTrees$new()
Creates a new instance of this R6 class.
Usage
AutoExtraTrees$new(id = "extra_trees")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoExtraTrees$graph()
Create the graph for the auto.
Usage
AutoExtraTrees$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
numeric(1)). timeout-
(
numeric(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoExtraTrees$estimate_memory()
Estimate the memory for the auto.
Usage
AutoExtraTrees$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoExtraTrees$clone()
The objects of this class are cloneable with this method.
Usage
AutoExtraTrees$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("extra_trees")auto("extra_trees")
Fastai Auto
Description
Fastai auto.
Value
Object of class R6::R6Class and AutoFastai.
Python learners
Python learners like TabPFN, TabFM, and fastai run via reticulate and
therefore need a Python installation with their required packages.
There are two ways to provide it:
-
Do nothing and let
reticulate::py_require()install the required packages into an ephemeral virtual environment automatically. -
Point the
RETICULATE_PYTHONenvironment variable to a Python installation that has the required packages installed.
We recommend option 2 when running on many workers,
as it avoids the overhead of downloading and installing the packages on each worker.
Use install_python_learners() to create a conda environment with the required packages and set
RETICULATE_PYTHON to the returned Python binary.
The TabPFN learner additionally requires the TABPFN_TOKEN environment variable to download the model weights.
Super class
Auto -> AutoFastai
Methods
Public methods
Inherited methods
AutoFastai$new()
Creates a new instance of this R6 class.
Usage
AutoFastai$new(id = "fastai")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoFastai$check()
Check if the auto is compatible with the task.
Usage
AutoFastai$check( task, memory_limit = Inf, large_data_set = FALSE, devices = "cpu" )
Arguments
task-
(mlr3::Task).
memory_limit-
(
integer(1)). large_data_set-
(
logical(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoFastai$graph()
Create the graph for the auto.
Usage
AutoFastai$graph( task, measure, n_threads, timeout, devices, isolate_python = TRUE )
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu". isolate_python-
(
logical(1))
Whether to train and predict the learner in a freshcallrsession. Isolation is required whenmlr3torchis loaded in the same process, because the bundled libtorch versions of the R and Python torch packages are incompatible. Default isTRUE.
AutoFastai$estimate_memory()
Estimate the memory for the auto.
Usage
AutoFastai$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoFastai$internal_measure()
Get the internal measure for the auto.
Usage
AutoFastai$internal_measure(measure, task)
Arguments
measuretask-
(mlr3::Task).
AutoFastai$clone()
The objects of this class are cloneable with this method.
Usage
AutoFastai$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("fastai")auto("fastai")
FTTransformer Auto
Description
FTTransformer auto.
Value
Object of class R6::R6Class and AutoFTTransformer.
Super class
Auto -> AutoFTTransformer
Methods
Public methods
Inherited methods
AutoFTTransformer$new()
Creates a new instance of this R6 class.
Usage
AutoFTTransformer$new(id = "ft_transformer")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoFTTransformer$graph()
Create the graph for the auto.
Usage
AutoFTTransformer$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoFTTransformer$estimate_memory()
Estimate the memory for the auto.
Usage
AutoFTTransformer$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoFTTransformer$clone()
The objects of this class are cloneable with this method.
Usage
AutoFTTransformer$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("ft_transformer")auto("ft_transformer")
Glmnet Auto
Description
Glmnet auto.
Value
Object of class R6::R6Class and AutoGlmnet.
Super class
Auto -> AutoGlmnet
Methods
Public methods
Inherited methods
AutoGlmnet$new()
Creates a new instance of this R6 class.
Usage
AutoGlmnet$new(id = "glmnet")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoGlmnet$graph()
Create the graph for the auto.
Usage
AutoGlmnet$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoGlmnet$clone()
The objects of this class are cloneable with this method.
Usage
AutoGlmnet$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("glmnet")auto("glmnet")
KKNN Auto
Description
Kknn auto.
Value
Object of class R6::R6Class and AutoKKNN.
Super class
Auto -> AutoKKNN
Methods
Public methods
Inherited methods
AutoKKNN$new()
Creates a new instance of this R6 class.
Usage
AutoKKNN$new(id = "kknn")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoKKNN$graph()
Create the graph for the auto.
Usage
AutoKKNN$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoKKNN$search_space()
Get the search space for the auto.
Usage
AutoKKNN$search_space(task)
Arguments
task-
(mlr3::Task).
AutoKKNN$clone()
The objects of this class are cloneable with this method.
Usage
AutoKKNN$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("kknn")auto("kknn")
Lda Auto
Description
Lda auto.
Value
Object of class R6::R6Class and AutoLda.
Super class
Auto -> AutoLda
Methods
Public methods
Inherited methods
AutoLda$new()
Creates a new instance of this R6 class.
Usage
AutoLda$new(id = "lda")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoLda$graph()
Create the graph for the auto.
Usage
AutoLda$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoLda$clone()
The objects of this class are cloneable with this method.
Usage
AutoLda$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("lda")auto("lda")
LightGBM Auto
Description
Lightgbm auto.
Value
Object of class R6::R6Class and AutoLightGBM.
Super class
Auto -> AutoLightGBM
Methods
Public methods
Inherited methods
AutoLightGBM$new()
Creates a new instance of this R6 class.
Usage
AutoLightGBM$new(id = "lightgbm")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoLightGBM$graph()
Create the graph for the auto.
Usage
AutoLightGBM$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoLightGBM$finalize_model()
Prepare the graph learner for the final model fit.
Usage
AutoLightGBM$finalize_model(graph_learner)
Arguments
graph_learner
AutoLightGBM$estimate_memory()
Estimate the memory for the auto.
Usage
AutoLightGBM$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoLightGBM$internal_measure()
Get the internal measure for the auto.
Usage
AutoLightGBM$internal_measure(measure, task)
Arguments
measuretask-
(mlr3::Task).
AutoLightGBM$clone()
The objects of this class are cloneable with this method.
Usage
AutoLightGBM$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("lightgbm")auto("lightgbm")
MLP Auto
Description
Mlp auto.
Value
Object of class R6::R6Class and AutoMLP.
Super class
Auto -> AutoMLP
Methods
Public methods
Inherited methods
AutoMLP$new()
Creates a new instance of this R6 class.
Usage
AutoMLP$new(id = "mlp")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoMLP$graph()
Create the graph for the auto.
Usage
AutoMLP$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoMLP$estimate_memory()
Estimate the memory for the auto.
Usage
AutoMLP$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoMLP$clone()
The objects of this class are cloneable with this method.
Usage
AutoMLP$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("mlp")auto("mlp")
Ranger Auto
Description
Ranger auto.
Value
Object of class R6::R6Class and AutoRanger.
Super class
Auto -> AutoRanger
Methods
Public methods
Inherited methods
AutoRanger$new()
Creates a new instance of this R6 class.
Usage
AutoRanger$new(id = "ranger")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoRanger$graph()
Create the graph for the auto.
Usage
AutoRanger$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoRanger$estimate_memory()
Estimate the memory for the auto.
Usage
AutoRanger$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoRanger$clone()
The objects of this class are cloneable with this method.
Usage
AutoRanger$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("ranger")auto("ranger")
ResNet Auto
Description
ResNet auto.
Value
Object of class R6::R6Class and AutoResNet.
Super class
Auto -> AutoResNet
Methods
Public methods
Inherited methods
AutoResNet$new()
Creates a new instance of this R6 class.
Usage
AutoResNet$new(id = "resnet")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoResNet$graph()
Create the graph for the auto.
Usage
AutoResNet$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoResNet$estimate_memory()
Estimate the memory for the auto.
Usage
AutoResNet$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoResNet$clone()
The objects of this class are cloneable with this method.
Usage
AutoResNet$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("resnet")auto("resnet")
SVM Auto
Description
Svm auto.
Value
Object of class R6::R6Class and AutoSVM.
Super class
Auto -> AutoSVM
Methods
Public methods
Inherited methods
AutoSVM$new()
Creates a new instance of this R6 class.
Usage
AutoSVM$new(id = "svm")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoSVM$graph()
Create the graph for the auto.
Usage
AutoSVM$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoSVM$design_default()
Default hyperparameters for the learner.
Usage
AutoSVM$design_default(task)
Arguments
task-
(mlr3::Task).
AutoSVM$clone()
The objects of this class are cloneable with this method.
Usage
AutoSVM$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("svm")auto("svm")
TabFM Auto
Description
Tabfm auto.
Tabfm predicts in context, so every prediction runs the backbone over the training rows once per estimator.
This is too slow to be useful on the CPU, so the auto is registered for "cuda" only and
Auto$check() removes it from the search space whenever devices does not include "cuda".
Construct it with AutoTabFM$new(devices = c("cpu", "cuda")) and re-register it in mlr_auto to run it
on the CPU anyway.
Value
Object of class R6::R6Class and AutoTabFM.
Python learners
Python learners like TabPFN, TabFM, and fastai run via reticulate and
therefore need a Python installation with their required packages.
There are two ways to provide it:
-
Do nothing and let
reticulate::py_require()install the required packages into an ephemeral virtual environment automatically. -
Point the
RETICULATE_PYTHONenvironment variable to a Python installation that has the required packages installed.
We recommend option 2 when running on many workers,
as it avoids the overhead of downloading and installing the packages on each worker.
Use install_python_learners() to create a conda environment with the required packages and set
RETICULATE_PYTHON to the returned Python binary.
The TabPFN learner additionally requires the TABPFN_TOKEN environment variable to download the model weights.
Super class
Auto -> AutoTabFM
Methods
Public methods
Inherited methods
AutoTabFM$new()
Creates a new instance of this R6 class.
Usage
AutoTabFM$new(id = "tabfm", devices = "cuda")
Arguments
id-
(
character(1))
Identifier for the new instance. devices-
(
character())
Devices the auto is allowed to run on. Defaults to"cuda"only, because tabfm is too slow to be useful on the CPU.
AutoTabFM$check()
Check if the auto is compatible with the task.
Usage
AutoTabFM$check( task, memory_limit = Inf, large_data_set = FALSE, devices = "cpu" )
Arguments
task-
(mlr3::Task).
memory_limit-
(
integer(1)). large_data_set-
(
logical(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoTabFM$graph()
Create the graph for the auto.
Usage
AutoTabFM$graph( task, measure, n_threads, timeout, devices, isolate_python = TRUE )
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu". isolate_python-
(
logical(1))
Whether to train and predict the learner in a freshcallrsession. Isolation is required whenmlr3torchis loaded in the same process, because the bundled libtorch versions of the R and Python torch packages are incompatible. Default isTRUE.
AutoTabFM$estimate_memory()
Estimate the memory for the auto.
Usage
AutoTabFM$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoTabFM$design_default()
Default hyperparameters for the learner.
Usage
AutoTabFM$design_default(task)
Arguments
task-
(mlr3::Task).
AutoTabFM$search_space()
Get the search space for the auto.
Usage
AutoTabFM$search_space(task)
Arguments
task-
(mlr3::Task).
AutoTabFM$clone()
The objects of this class are cloneable with this method.
Usage
AutoTabFM$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("tabfm")auto("tabfm")
TabPFN Auto
Description
Tabpfn auto.
Value
Object of class R6::R6Class and AutoTabPFN.
Python learners
Python learners like TabPFN, TabFM, and fastai run via reticulate and
therefore need a Python installation with their required packages.
There are two ways to provide it:
-
Do nothing and let
reticulate::py_require()install the required packages into an ephemeral virtual environment automatically. -
Point the
RETICULATE_PYTHONenvironment variable to a Python installation that has the required packages installed.
We recommend option 2 when running on many workers,
as it avoids the overhead of downloading and installing the packages on each worker.
Use install_python_learners() to create a conda environment with the required packages and set
RETICULATE_PYTHON to the returned Python binary.
The TabPFN learner additionally requires the TABPFN_TOKEN environment variable to download the model weights.
Super class
Auto -> AutoTabPFN
Methods
Public methods
Inherited methods
AutoTabPFN$new()
Creates a new instance of this R6 class.
Usage
AutoTabPFN$new(id = "tabpfn")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoTabPFN$check()
Check if the auto is compatible with the task.
Usage
AutoTabPFN$check( task, memory_limit = Inf, large_data_set = FALSE, devices = "cpu" )
Arguments
task-
(mlr3::Task).
memory_limit-
(
integer(1)). large_data_set-
(
logical(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoTabPFN$graph()
Create the graph for the auto.
Usage
AutoTabPFN$graph( task, measure, n_threads, timeout, devices, isolate_python = TRUE )
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu". isolate_python-
(
logical(1))
Whether to train and predict the learner in a freshcallrsession. Isolation is required whenmlr3torchis loaded in the same process, because the bundled libtorch versions of the R and Python torch packages are incompatible. Default isTRUE.
AutoTabPFN$estimate_memory()
Estimate the memory for the auto.
Usage
AutoTabPFN$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoTabPFN$design_default()
Default hyperparameters for the learner.
Usage
AutoTabPFN$design_default(task)
Arguments
task-
(mlr3::Task).
AutoTabPFN$search_space()
Get the search space for the auto.
Usage
AutoTabPFN$search_space(task)
Arguments
task-
(mlr3::Task).
AutoTabPFN$clone()
The objects of this class are cloneable with this method.
Usage
AutoTabPFN$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("tabpfn")auto("tabpfn")
Xgboost Auto
Description
Xgboost auto.
Value
Object of class R6::R6Class and AutoXgboost.
Super class
Auto -> AutoXgboost
Methods
Public methods
Inherited methods
AutoXgboost$new()
Creates a new instance of this R6 class.
Usage
AutoXgboost$new(id = "xgboost")
Arguments
id-
(
character(1))
Identifier for the new instance.
AutoXgboost$graph()
Create the graph for the auto.
Usage
AutoXgboost$graph(task, measure, n_threads, timeout, devices)
Arguments
task-
(mlr3::Task).
measuren_threads-
(
integer(1)). timeout-
(
integer(1)). devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoXgboost$finalize_model()
Prepare the graph learner for the final model fit.
Usage
AutoXgboost$finalize_model(graph_learner)
Arguments
graph_learner
AutoXgboost$estimate_memory()
Estimate the memory for the auto.
Usage
AutoXgboost$estimate_memory(task, devices = "cpu")
Arguments
task-
(mlr3::Task).
devices-
(
character())
Devices to use. Allowed values are"cpu"and"cuda". Default is "cpu".
AutoXgboost$internal_measure()
Get the internal measure for the auto.
Usage
AutoXgboost$internal_measure(measure, task)
Arguments
measuretask-
(mlr3::Task).
AutoXgboost$clone()
The objects of this class are cloneable with this method.
Usage
AutoXgboost$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
auto("xgboost")auto("xgboost")
Install Python Learners
Description
Creates a conda environment and installs the packages required by the Python learners.
Installs the dependencies of the fastai auto learner, the tabfm auto learner, and the tabpfn auto learner.
The learners share a single environment, so one RETICULATE_PYTHON covers all of them.
Usage
install_python_learners( learners = c("fastai", "tabfm", "tabpfn"), envname = file.path(tools::R_user_dir("mlr3automl", which = "data"), "mlr3automl-python"), python_version = "3.12" )install_python_learners( learners = c("fastai", "tabfm", "tabpfn"), envname = file.path(tools::R_user_dir("mlr3automl", which = "data"), "mlr3automl-python"), python_version = "3.12" )
Arguments
learners |
( |
envname |
( |
python_version |
( |
Details
The environment is created under tools::R_user_dir() by default.
The environment can exceed 10GB with torch/CUDA dependencies.
The function downloads and installs Python packages and must therefore be called explicitly by the user.
It is never called by any other function of the package.
Value
Invisibly returns the path to the Python binary in the environment.
Examples
# The example is wrapped in \dontrun{} because it cannot be executed during checks. # It downloads and installs conda, a Python interpreter, and several Python packages, # which requires network access, several GB of disk space, and a few minutes of runtime. ## Not run: python = install_python_learners() Sys.setenv(RETICULATE_PYTHON = python) ## End(Not run)# The example is wrapped in \dontrun{} because it cannot be executed during checks. # It downloads and installs conda, a Python interpreter, and several Python packages, # which requires network access, several GB of disk space, and a few minutes of runtime. ## Not run: python = install_python_learners() Sys.setenv(RETICULATE_PYTHON = python) ## End(Not run)
Classification AutoML Learner
Description
The LearnerClassifAuto is an automated machine learning (AutoML) system for classification tasks. It combines preprocessing, a switch between multiple learners, and hyperparameter tuning to find the best model for the given task.
Value
Object of class R6::R6Class and LearnerClassifAuto.
Debugging
Set options(bbotk.debug = TRUE) to run the tuning in the main session.
Set encapsulate_learner = FALSE to remove encapsulation of the learner.
Set encapsulate_mbo = FALSE to catch no errors in mbo.
Parameters
- learner_timeout
-
(
integer(1))
Timeout for training and predicting with a single learner. - n_threads
-
(
integer(1))
Number of threads used for training a single learner. - n_cpu
-
(named
integer())
Number of CPUs a single training of a learner uses, named by learner id, e.g.c(xgboost = 1). Overrides the default of the learner. Must be at least 1 because the worker always runs on the CPU. Currently informational only; the number of threads is controlled byn_threads. - n_gpu
-
(named
integer())
Number of GPUs a single training of a learner uses, named by learner id, e.g.c(xgboost = 1). Overrides the default of the learner. Can only be 0 or 1 for now. The torch learners, TabPFN, TabFM, and fastai default to 1; all other learners default to 0. Only effective when"cuda"is part ofdevices; otherwise every learner is trained on the CPU. When the requirements are mixed and the daemons of the mirai compute profiles"mlr3automl_cpu"and"mlr3automl_gpu"are set up withrush::rush_plan(), the search space is partitioned into a cpu and a gpu subspace which are tuned with mlr3mbo::TunerADBOSubspaces. The workers of a profile only ever propose and evaluate points of the subspace of that profile, so the number of workers per subspace is the number of workers of its profile. Otherwise the cpu and gpu learners are tuned in a single search space with mlr3mbo::TunerAsyncMbo.mirai::daemons(7, .compute = "mlr3automl_cpu") mirai::daemons(1, .compute = "mlr3automl_gpu") rush::rush_plan(profiles = c(mlr3automl_cpu = 7, mlr3automl_gpu = 1))
- memory_limit
-
(
integer(1))
Memory limit for training a single learner in MB. The limit is shared across the parallel workers, i.e. divided by the number of workers. - devices
-
(
character())
Devices to use for model training. Possible values are"cpu"and"cuda". If"cuda", learners with an_gpurequirement of 1 are trained on a GPU, while the remaining learners stay on the CPU. - large_data_size
-
(
integer(1))
Threshold for the data set size (number of rows times number of columns) above which large-data rules apply. Beyond this threshold the number of parallel workers is reduced to a quarter, rounded up, and each remaining worker is given proportionally more threads and memory. When the workers are distributed over mirai compute profiles, the number of workers of every profile is reduced, but every profile keeps at least one worker. The"mlr3automl_gpu"profile is exempt because its number of workers is fixed by the number of GPUs. It keeps its workers, threads, and memory limit so that the gpu learners do not claim the CPU cores and the memory that are freed on the cpu profiles. - small_data_size
-
(
integer(1))
Threshold value for the data set size (rows) from which special rules apply. - small_data_resampling
-
(mlr3::Resampling)
Resampling strategy to use for model training on small data sets. - initial_design_default
-
(
logical(1))
Whether to use the default design of the learner. - initial_design_set
-
(
integer(1))
Number of points to use for the initial design set. - initial_design_size
-
(
integer(1))
Size of the random, sobol or lhs initial design. - initial_design_type
-
(
character(1))
Type of the initial design used for mbo. Possible values are"lhs","sobol","random"."lhs"uses a Latin Hypercube Sampling design."sobol"uses a Sobol sequence design."random"uses a random design. - initial_design_fraction
-
(
numeric(1))
Fraction of the budget to use for the initial design. When the search space is partitioned into a cpu and a gpu subspace, the remaining points of both designs are dropped, because every compute profile has its own queue. - resampling
-
(mlr3::Resampling)
Resampling strategy used for tuning. - terminator
-
(bbotk::Terminator)
Terminator criterion for tuning. - measure
-
(mlr3::Measure)
Measure used for tuning. - callbacks
-
(mlr3tuning::CallbackAsyncTuning)
Callbacks used for tuning. - store_benchmark_result
-
(
logical(1))
Whether to store the benchmark result. - store_models
-
(
logical(1))
Whether to store the models. - encapsulate_learner
-
(
logical(1))
Whether to encapsulate the learner. Change toFALSEto debug. - encapsulate_mbo
-
(
logical(1))
Whether to encapsulate the tuning. Change toFALSEto debug. - check_learners
-
(
logical(1))
Whether to check if the learners are compatible with the task. Change toFALSEto debug.
Python learners
Python learners like TabPFN, TabFM, and fastai run via reticulate and
therefore need a Python installation with their required packages.
There are two ways to provide it:
-
Do nothing and let
reticulate::py_require()install the required packages into an ephemeral virtual environment automatically. -
Point the
RETICULATE_PYTHONenvironment variable to a Python installation that has the required packages installed.
We recommend option 2 when running on many workers,
as it avoids the overhead of downloading and installing the packages on each worker.
Use install_python_learners() to create a conda environment with the required packages and set
RETICULATE_PYTHON to the returned Python binary.
The TabPFN learner additionally requires the TABPFN_TOKEN environment variable to download the model weights.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto
Methods
Public methods
Inherited methods
LearnerClassifAuto$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAuto$new(id = "classif.auto", learner_ids, rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. learner_ids-
(
character())
Learner that should be used. rush-
rush::Rush
Rush instance.
LearnerClassifAuto$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAuto$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
packages = c( "mlr3extralearners", "catboost", "ranger", "callr", "mlr3torch", "glmnet", "kknn", "MASS", "lightgbm", "e1071", "xgboost" ) if (mlr3misc::require_namespaces(packages, quietly = TRUE)) { learner = lrn("classif.auto") learner }packages = c( "mlr3extralearners", "catboost", "ranger", "callr", "mlr3torch", "glmnet", "kknn", "MASS", "lightgbm", "e1071", "xgboost" ) if (mlr3misc::require_namespaces(packages, quietly = TRUE)) { learner = lrn("classif.auto") learner }
Classification Gradient Boosted Decision Trees Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoCatboost.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoCatboost
Methods
Public methods
Inherited methods
LearnerClassifAutoCatboost$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoCatboost$new(id = "classif.auto_catboost", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoCatboost$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoCatboost$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces(c("mlr3extralearners", "catboost"), quietly = TRUE)) { learner = lrn("classif.auto_catboost") learner }if (mlr3misc::require_namespaces(c("mlr3extralearners", "catboost"), quietly = TRUE)) { learner = lrn("classif.auto_catboost") learner }
Classification Fastai Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoFastai.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoFastai
Methods
Public methods
Inherited methods
LearnerClassifAutoFastai$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoFastai$new(id = "classif.auto_fastai", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoFastai$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoFastai$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces(c("mlr3extralearners", "callr"), quietly = TRUE)) { learner = lrn("classif.auto_fastai") learner }if (mlr3misc::require_namespaces(c("mlr3extralearners", "callr"), quietly = TRUE)) { learner = lrn("classif.auto_fastai") learner }
Classification FT-Transformer Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoFTTransformer.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoFTTransformer
Methods
Public methods
Inherited methods
LearnerClassifAutoFTTransformer$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoFTTransformer$new( id = "classif.auto_ft_transformer", rush = NULL )
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoFTTransformer$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoFTTransformer$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("classif.auto_ft_transformer") learner }if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("classif.auto_ft_transformer") learner }
Classification GLM with Elastic Net Regularization Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoGlmnet.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoGlmnet
Methods
Public methods
Inherited methods
LearnerClassifAutoGlmnet$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoGlmnet$new(id = "classif.auto_glmnet", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoGlmnet$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoGlmnet$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("glmnet", quietly = TRUE)) { learner = lrn("classif.auto_glmnet") learner }if (mlr3misc::require_namespaces("glmnet", quietly = TRUE)) { learner = lrn("classif.auto_glmnet") learner }
Classification k-Nearest-Neighbor Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoKKNN.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoKKNN
Methods
Public methods
Inherited methods
LearnerClassifAutoKKNN$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoKKNN$new(id = "classif.auto_kknn", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoKKNN$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoKKNN$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("kknn", quietly = TRUE)) { learner = lrn("classif.auto_kknn") learner }if (mlr3misc::require_namespaces("kknn", quietly = TRUE)) { learner = lrn("classif.auto_kknn") learner }
Classification LightGBM Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoLightGBM.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoLightGBM
Methods
Public methods
Inherited methods
LearnerClassifAutoLightGBM$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoLightGBM$new(id = "classif.auto_lightgbm", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoLightGBM$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoLightGBM$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces(c("mlr3extralearners", "lightgbm"), quietly = TRUE)) { learner = lrn("classif.auto_lightgbm") learner }if (mlr3misc::require_namespaces(c("mlr3extralearners", "lightgbm"), quietly = TRUE)) { learner = lrn("classif.auto_lightgbm") learner }
Classification MLP Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoMLP.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoMLP
Methods
Public methods
Inherited methods
LearnerClassifAutoMLP$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoMLP$new(id = "classif.auto_mlp", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoMLP$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoMLP$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("classif.auto_mlp") learner }if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("classif.auto_mlp") learner }
Classification Ranger Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoRanger.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoRanger
Methods
Public methods
Inherited methods
LearnerClassifAutoRanger$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoRanger$new(id = "classif.auto_ranger", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoRanger$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoRanger$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("ranger", quietly = TRUE)) { learner = lrn("classif.auto_ranger") learner }if (mlr3misc::require_namespaces("ranger", quietly = TRUE)) { learner = lrn("classif.auto_ranger") learner }
Classification ResNet Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoResNet.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoResNet
Methods
Public methods
Inherited methods
LearnerClassifAutoResNet$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoResNet$new(id = "classif.auto_resnet", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoResNet$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoResNet$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("classif.auto_resnet") learner }if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("classif.auto_resnet") learner }
Classification SVM Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoSVM.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoSVM
Methods
Public methods
Inherited methods
LearnerClassifAutoSVM$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoSVM$new(id = "classif.auto_svm", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoSVM$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoSVM$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("e1071", quietly = TRUE)) { learner = lrn("classif.auto_svm") learner }if (mlr3misc::require_namespaces("e1071", quietly = TRUE)) { learner = lrn("classif.auto_svm") learner }
Classification TabFM Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoTabFM.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoTabFM
Methods
Public methods
Inherited methods
LearnerClassifAutoTabFM$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoTabFM$new(id = "classif.auto_tabfm", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoTabFM$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoTabFM$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces(c("mlr3extralearners", "callr"), quietly = TRUE)) { learner = lrn("classif.auto_tabfm") learner }if (mlr3misc::require_namespaces(c("mlr3extralearners", "callr"), quietly = TRUE)) { learner = lrn("classif.auto_tabfm") learner }
Classification TabPFN Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoTabPFN.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoTabPFN
Methods
Public methods
Inherited methods
LearnerClassifAutoTabPFN$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoTabPFN$new(id = "classif.auto_tabpfn", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoTabPFN$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoTabPFN$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces(c("mlr3extralearners", "callr"), quietly = TRUE)) { learner = lrn("classif.auto_tabpfn") learner }if (mlr3misc::require_namespaces(c("mlr3extralearners", "callr"), quietly = TRUE)) { learner = lrn("classif.auto_tabpfn") learner }
Classification XGBoost Auto Learner
Description
Classification auto learner.
Value
Object of class R6::R6Class and LearnerClassifAutoXgboost.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerClassifAuto -> LearnerClassifAutoXgboost
Methods
Public methods
Inherited methods
LearnerClassifAutoXgboost$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifAutoXgboost$new(id = "classif.auto_xgboost", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerClassifAutoXgboost$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifAutoXgboost$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("xgboost", quietly = TRUE)) { learner = lrn("classif.auto_xgboost") learner }if (mlr3misc::require_namespaces("xgboost", quietly = TRUE)) { learner = lrn("classif.auto_xgboost") learner }
Fastai Learner Isolated
Description
A subclass of mlr3extralearners::LearnerClassifFastai that isolates the Python environment in a callr session.
Value
Object of class R6::R6Class and LearnerClassifFastaiIsolated.
Super classes
mlr3::Learner -> mlr3::LearnerClassif -> mlr3extralearners::LearnerClassifFastai -> LearnerClassifFastaiIsolated
Public fields
isolate_python-
(
logical(1))
Whether to run.train()and.predict()in a fresh callr session. Set by AutoFastai$graph(); onlyFALSEwhen the run's learners never load mlr3torch.
Methods
Public methods
Inherited methods
mlr3::Learner$base_learner()mlr3::Learner$configure()mlr3::Learner$encapsulate()mlr3::Learner$format()mlr3::Learner$help()mlr3::Learner$predict()mlr3::Learner$predict_newdata()mlr3::Learner$print()mlr3::Learner$reset()mlr3::Learner$selected_features()mlr3::Learner$train()mlr3::LearnerClassif$predict_newdata_fast()mlr3extralearners::LearnerClassifFastai$marshal()mlr3extralearners::LearnerClassifFastai$unmarshal()
LearnerClassifFastaiIsolated$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifFastaiIsolated$new()
LearnerClassifFastaiIsolated$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifFastaiIsolated$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
TabFM Learner Isolated
Description
A subclass of mlr3extralearners::LearnerClassifTabFM that isolates the Python environment in a callr session.
Value
Object of class R6::R6Class and LearnerClassifTabFMIsolated.
Super classes
mlr3::Learner -> mlr3::LearnerClassif -> mlr3extralearners::LearnerClassifTabFM -> LearnerClassifTabFMIsolated
Public fields
isolate_python-
(
logical(1))
Whether to run.train()and.predict()in a fresh callr session. Set by AutoTabFM$graph(); onlyFALSEwhen the run's learners never load mlr3torch.
Methods
Public methods
Inherited methods
mlr3::Learner$base_learner()mlr3::Learner$configure()mlr3::Learner$encapsulate()mlr3::Learner$format()mlr3::Learner$help()mlr3::Learner$predict()mlr3::Learner$predict_newdata()mlr3::Learner$print()mlr3::Learner$reset()mlr3::Learner$selected_features()mlr3::Learner$train()mlr3::LearnerClassif$predict_newdata_fast()mlr3extralearners::LearnerClassifTabFM$marshal()mlr3extralearners::LearnerClassifTabFM$unmarshal()
LearnerClassifTabFMIsolated$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifTabFMIsolated$new()
LearnerClassifTabFMIsolated$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifTabFMIsolated$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
TabPFN Learner Isolated
Description
A subclass of mlr3extralearners::LearnerClassifTabPFN that isolates the Python environment in a callr session.
Value
Object of class R6::R6Class and LearnerClassifTabPFNIsolated.
Super classes
mlr3::Learner -> mlr3::LearnerClassif -> mlr3extralearners::LearnerClassifTabPFN -> LearnerClassifTabPFNIsolated
Public fields
isolate_python-
(
logical(1))
Whether to run.train()and.predict()in a fresh callr session. Set by AutoTabPFN$graph(); onlyFALSEwhen the run's learners never load mlr3torch.
Methods
Public methods
Inherited methods
mlr3::Learner$base_learner()mlr3::Learner$configure()mlr3::Learner$encapsulate()mlr3::Learner$format()mlr3::Learner$help()mlr3::Learner$predict()mlr3::Learner$predict_newdata()mlr3::Learner$print()mlr3::Learner$reset()mlr3::Learner$selected_features()mlr3::Learner$train()mlr3::LearnerClassif$predict_newdata_fast()mlr3extralearners::LearnerClassifTabPFN$marshal()mlr3extralearners::LearnerClassifTabPFN$unmarshal()
LearnerClassifTabPFNIsolated$new()
Creates a new instance of this R6 class.
Usage
LearnerClassifTabPFNIsolated$new()
LearnerClassifTabPFNIsolated$clone()
The objects of this class are cloneable with this method.
Usage
LearnerClassifTabPFNIsolated$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Regression AutoML Learner
Description
The LearnerRegrAuto is an automated machine learning (AutoML) system for regression tasks. It combines preprocessing, a switch between multiple learners, and hyperparameter tuning to find the best model for the given task.
Value
Object of class R6::R6Class and LearnerRegrAuto.
Debugging
Set options(bbotk.debug = TRUE) to run the tuning in the main session.
Set encapsulate_learner = FALSE to remove encapsulation of the learner.
Set encapsulate_mbo = FALSE to catch no errors in mbo.
Parameters
- learner_timeout
-
(
integer(1))
Timeout for training and predicting with a single learner. - n_threads
-
(
integer(1))
Number of threads used for training a single learner. - n_cpu
-
(named
integer())
Number of CPUs a single training of a learner uses, named by learner id, e.g.c(xgboost = 1). Overrides the default of the learner. Must be at least 1 because the worker always runs on the CPU. Currently informational only; the number of threads is controlled byn_threads. - n_gpu
-
(named
integer())
Number of GPUs a single training of a learner uses, named by learner id, e.g.c(xgboost = 1). Overrides the default of the learner. Can only be 0 or 1 for now. The torch learners, TabPFN, TabFM, and fastai default to 1; all other learners default to 0. Only effective when"cuda"is part ofdevices; otherwise every learner is trained on the CPU. When the requirements are mixed and the daemons of the mirai compute profiles"mlr3automl_cpu"and"mlr3automl_gpu"are set up withrush::rush_plan(), the search space is partitioned into a cpu and a gpu subspace which are tuned with mlr3mbo::TunerADBOSubspaces. The workers of a profile only ever propose and evaluate points of the subspace of that profile, so the number of workers per subspace is the number of workers of its profile. Otherwise the cpu and gpu learners are tuned in a single search space with mlr3mbo::TunerAsyncMbo.mirai::daemons(7, .compute = "mlr3automl_cpu") mirai::daemons(1, .compute = "mlr3automl_gpu") rush::rush_plan(profiles = c(mlr3automl_cpu = 7, mlr3automl_gpu = 1))
- memory_limit
-
(
integer(1))
Memory limit for training a single learner in MB. The limit is shared across the parallel workers, i.e. divided by the number of workers. - devices
-
(
character())
Devices to use for model training. Possible values are"cpu"and"cuda". If"cuda", learners with an_gpurequirement of 1 are trained on a GPU, while the remaining learners stay on the CPU. - large_data_size
-
(
integer(1))
Threshold for the data set size (number of rows times number of columns) above which large-data rules apply. Beyond this threshold the number of parallel workers is reduced to a quarter, rounded up, and each remaining worker is given proportionally more threads and memory. When the workers are distributed over mirai compute profiles, the number of workers of every profile is reduced, but every profile keeps at least one worker. The"mlr3automl_gpu"profile is exempt because its number of workers is fixed by the number of GPUs. It keeps its workers, threads, and memory limit so that the gpu learners do not claim the CPU cores and the memory that are freed on the cpu profiles. - small_data_size
-
(
integer(1))
Threshold value for the data set size (rows) from which special rules apply. - small_data_resampling
-
(mlr3::Resampling)
Resampling strategy to use for model training on small data sets. - initial_design_default
-
(
logical(1))
Whether to use the default design of the learner. - initial_design_set
-
(
integer(1))
Number of points to use for the initial design set. - initial_design_size
-
(
integer(1))
Size of the random, sobol or lhs initial design. - initial_design_type
-
(
character(1))
Type of the initial design used for mbo. Possible values are"lhs","sobol","random"."lhs"uses a Latin Hypercube Sampling design."sobol"uses a Sobol sequence design."random"uses a random design. - initial_design_fraction
-
(
numeric(1))
Fraction of the budget to use for the initial design. When the search space is partitioned into a cpu and a gpu subspace, the remaining points of both designs are dropped, because every compute profile has its own queue. - resampling
-
(mlr3::Resampling)
Resampling strategy used for tuning. - terminator
-
(bbotk::Terminator)
Terminator criterion for tuning. - measure
-
(mlr3::Measure)
Measure used for tuning. - callbacks
-
(mlr3tuning::CallbackAsyncTuning)
Callbacks used for tuning. - store_benchmark_result
-
(
logical(1))
Whether to store the benchmark result. - store_models
-
(
logical(1))
Whether to store the models. - encapsulate_learner
-
(
logical(1))
Whether to encapsulate the learner. Change toFALSEto debug. - encapsulate_mbo
-
(
logical(1))
Whether to encapsulate the tuning. Change toFALSEto debug. - check_learners
-
(
logical(1))
Whether to check if the learners are compatible with the task. Change toFALSEto debug.
Python learners
Python learners like TabPFN, TabFM, and fastai run via reticulate and
therefore need a Python installation with their required packages.
There are two ways to provide it:
-
Do nothing and let
reticulate::py_require()install the required packages into an ephemeral virtual environment automatically. -
Point the
RETICULATE_PYTHONenvironment variable to a Python installation that has the required packages installed.
We recommend option 2 when running on many workers,
as it avoids the overhead of downloading and installing the packages on each worker.
Use install_python_learners() to create a conda environment with the required packages and set
RETICULATE_PYTHON to the returned Python binary.
The TabPFN learner additionally requires the TABPFN_TOKEN environment variable to download the model weights.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto
Methods
Public methods
Inherited methods
LearnerRegrAuto$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAuto$new(id = "regr.auto", learner_ids, rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. learner_ids-
(
character())
Learner that should be used. rush-
rush::Rush
Rush instance.
LearnerRegrAuto$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAuto$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
packages = c( "mlr3extralearners", "catboost", "ranger", "callr", "mlr3torch", "glmnet", "kknn", "MASS", "lightgbm", "e1071", "xgboost" ) if (mlr3misc::require_namespaces(packages, quietly = TRUE)) { learner = lrn("regr.auto") learner }packages = c( "mlr3extralearners", "catboost", "ranger", "callr", "mlr3torch", "glmnet", "kknn", "MASS", "lightgbm", "e1071", "xgboost" ) if (mlr3misc::require_namespaces(packages, quietly = TRUE)) { learner = lrn("regr.auto") learner }
Regression Gradient Boosted Decision Trees Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoCatboost.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoCatboost
Methods
Public methods
Inherited methods
LearnerRegrAutoCatboost$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoCatboost$new(id = "regr.auto_catboost", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoCatboost$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoCatboost$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces(c("mlr3extralearners", "catboost"), quietly = TRUE)) { learner = lrn("regr.auto_catboost") learner }if (mlr3misc::require_namespaces(c("mlr3extralearners", "catboost"), quietly = TRUE)) { learner = lrn("regr.auto_catboost") learner }
Regression FT-Transformer Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoFTTransformer.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoFTTransformer
Methods
Public methods
Inherited methods
LearnerRegrAutoFTTransformer$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoFTTransformer$new(id = "regr.auto_ft_transformer", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoFTTransformer$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoFTTransformer$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("regr.auto_ft_transformer") learner }if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("regr.auto_ft_transformer") learner }
Regression GLM with Elastic Net Regularization Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoGlmnet.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoGlmnet
Methods
Public methods
Inherited methods
LearnerRegrAutoGlmnet$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoGlmnet$new(id = "regr.auto_glmnet", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoGlmnet$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoGlmnet$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("glmnet", quietly = TRUE)) { learner = lrn("regr.auto_glmnet") learner }if (mlr3misc::require_namespaces("glmnet", quietly = TRUE)) { learner = lrn("regr.auto_glmnet") learner }
Regression k-Nearest-Neighbor Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoKKNN.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoKKNN
Methods
Public methods
Inherited methods
LearnerRegrAutoKKNN$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoKKNN$new(id = "regr.auto_kknn", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoKKNN$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoKKNN$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("kknn", quietly = TRUE)) { learner = lrn("regr.auto_kknn") learner }if (mlr3misc::require_namespaces("kknn", quietly = TRUE)) { learner = lrn("regr.auto_kknn") learner }
Regression LightGBM Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoLightGBM.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoLightGBM
Methods
Public methods
Inherited methods
LearnerRegrAutoLightGBM$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoLightGBM$new(id = "regr.auto_lightgbm", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoLightGBM$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoLightGBM$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces(c("mlr3extralearners", "lightgbm"), quietly = TRUE)) { learner = lrn("regr.auto_lightgbm") learner }if (mlr3misc::require_namespaces(c("mlr3extralearners", "lightgbm"), quietly = TRUE)) { learner = lrn("regr.auto_lightgbm") learner }
Regression MLP Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoMLP.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoMLP
Methods
Public methods
Inherited methods
LearnerRegrAutoMLP$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoMLP$new(id = "regr.auto_mlp", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoMLP$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoMLP$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("regr.auto_mlp") learner }if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("regr.auto_mlp") learner }
Regression Ranger Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoRanger.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoRanger
Methods
Public methods
Inherited methods
LearnerRegrAutoRanger$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoRanger$new(id = "regr.auto_ranger", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoRanger$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoRanger$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("ranger", quietly = TRUE)) { learner = lrn("regr.auto_ranger") learner }if (mlr3misc::require_namespaces("ranger", quietly = TRUE)) { learner = lrn("regr.auto_ranger") learner }
Regression ResNet Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoResNet.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoResNet
Methods
Public methods
Inherited methods
LearnerRegrAutoResNet$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoResNet$new(id = "regr.auto_resnet", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoResNet$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoResNet$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("regr.auto_resnet") learner }if (mlr3misc::require_namespaces("mlr3torch", quietly = TRUE)) { learner = lrn("regr.auto_resnet") learner }
Regression SVM Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoSVM.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoSVM
Methods
Public methods
Inherited methods
LearnerRegrAutoSVM$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoSVM$new(id = "regr.auto_svm", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoSVM$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoSVM$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("e1071", quietly = TRUE)) { learner = lrn("regr.auto_svm") learner }if (mlr3misc::require_namespaces("e1071", quietly = TRUE)) { learner = lrn("regr.auto_svm") learner }
Regression TabFM Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoTabFM.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoTabFM
Methods
Public methods
Inherited methods
LearnerRegrAutoTabFM$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoTabFM$new(id = "regr.auto_tabfm", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoTabFM$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoTabFM$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces(c("mlr3extralearners", "callr"), quietly = TRUE)) { learner = lrn("regr.auto_tabfm") learner }if (mlr3misc::require_namespaces(c("mlr3extralearners", "callr"), quietly = TRUE)) { learner = lrn("regr.auto_tabfm") learner }
Regression TabPFN Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoTabPFN.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoTabPFN
Methods
Public methods
Inherited methods
LearnerRegrAutoTabPFN$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoTabPFN$new(id = "regr.auto_tabpfn", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoTabPFN$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoTabPFN$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces(c("mlr3extralearners", "callr"), quietly = TRUE)) { learner = lrn("regr.auto_tabpfn") learner }if (mlr3misc::require_namespaces(c("mlr3extralearners", "callr"), quietly = TRUE)) { learner = lrn("regr.auto_tabpfn") learner }
Regression XGBoost Auto Learner
Description
Regression auto learner.
Value
Object of class R6::R6Class and LearnerRegrAutoXgboost.
Super classes
mlr3::Learner -> LearnerAuto -> LearnerRegrAuto -> LearnerRegrAutoXgboost
Methods
Public methods
Inherited methods
LearnerRegrAutoXgboost$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrAutoXgboost$new(id = "regr.auto_xgboost", rush = NULL)
Arguments
id-
(
character(1))
Identifier for the new instance. rush-
rush::Rush
Rush instance.
LearnerRegrAutoXgboost$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrAutoXgboost$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Examples
if (mlr3misc::require_namespaces("xgboost", quietly = TRUE)) { learner = lrn("regr.auto_xgboost") learner }if (mlr3misc::require_namespaces("xgboost", quietly = TRUE)) { learner = lrn("regr.auto_xgboost") learner }
TabFM Regressor Learner Isolated
Description
A subclass of mlr3extralearners::LearnerRegrTabFM that isolates the Python environment in a callr session.
Value
Object of class R6::R6Class and LearnerRegrTabFMIsolated.
Super classes
mlr3::Learner -> mlr3::LearnerRegr -> mlr3extralearners::LearnerRegrTabFM -> LearnerRegrTabFMIsolated
Public fields
isolate_python-
(
logical(1))
Whether to run.train()and.predict()in a fresh callr session. Set by AutoTabFM$graph(); onlyFALSEwhen the run's learners never load mlr3torch.
Methods
Public methods
Inherited methods
mlr3::Learner$base_learner()mlr3::Learner$configure()mlr3::Learner$encapsulate()mlr3::Learner$format()mlr3::Learner$help()mlr3::Learner$predict()mlr3::Learner$predict_newdata()mlr3::Learner$print()mlr3::Learner$reset()mlr3::Learner$selected_features()mlr3::Learner$train()mlr3::LearnerRegr$predict_newdata_fast()mlr3extralearners::LearnerRegrTabFM$marshal()mlr3extralearners::LearnerRegrTabFM$unmarshal()
LearnerRegrTabFMIsolated$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrTabFMIsolated$new()
LearnerRegrTabFMIsolated$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrTabFMIsolated$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
TabPFN Regressor Learner Isolated
Description
A subclass of mlr3extralearners::LearnerRegrTabPFN that isolates the Python environment in a callr session.
Value
Object of class R6::R6Class and LearnerRegrTabPFNIsolated.
Super classes
mlr3::Learner -> mlr3::LearnerRegr -> mlr3extralearners::LearnerRegrTabPFN -> LearnerRegrTabPFNIsolated
Public fields
isolate_python-
(
logical(1))
Whether to run.train()and.predict()in a fresh callr session. Set by AutoTabPFN$graph(); onlyFALSEwhen the run's learners never load mlr3torch.
Methods
Public methods
Inherited methods
mlr3::Learner$base_learner()mlr3::Learner$configure()mlr3::Learner$encapsulate()mlr3::Learner$format()mlr3::Learner$help()mlr3::Learner$predict()mlr3::Learner$predict_newdata()mlr3::Learner$print()mlr3::Learner$reset()mlr3::Learner$selected_features()mlr3::Learner$train()mlr3::LearnerRegr$predict_newdata_fast()mlr3extralearners::LearnerRegrTabPFN$marshal()mlr3extralearners::LearnerRegrTabPFN$unmarshal()
LearnerRegrTabPFNIsolated$new()
Creates a new instance of this R6 class.
Usage
LearnerRegrTabPFNIsolated$new()
LearnerRegrTabPFNIsolated$clone()
The objects of this class are cloneable with this method.
Usage
LearnerRegrTabPFNIsolated$clone(deep = FALSE)
Arguments
deep-
Whether to make a deep clone.
Dictionary of Auto Objects
Description
A dictionary of Auto objects.
Sugar function to retrieve Auto objects from mlr_auto.
Usage
mlr_auto auto(.key, ...)mlr_auto auto(.key, ...)
Arguments
.key |
( |
... |
(named |
Value
Examples
auto("catboost")auto("catboost")
Encapsulation Daemon Callback
Description
This mlr3misc::Callback starts a persistent mirai daemon on each rush worker that is reused for the
"mirai" encapsulation of the tuned learners.
Reusing a single daemon avoids the overhead of starting a new daemon for every learner evaluation.
The daemon is started when the worker begins and its liveness is checked before every evaluation.
If the daemon has died, it is restarted.
Parameters
-
n_daemons::integer(1)
Number of daemons to start on each worker. Defaults to1.
Examples
clbk("mlr3automl.encapsulation_daemon", n_daemons = 1)clbk("mlr3automl.encapsulation_daemon", n_daemons = 1)
Initial Design Runtime Limit Callback
Description
This mlr3misc::Callback drops the remaining tasks of the initial design from the queue when a
configurable fraction (initial_design_fraction) of the runtime limit is reached.
The initial_design_fraction is set to 0.25 by default.
Examples
clbk("mlr3automl.initial_design_runtime", initial_design_fraction = 0.5)clbk("mlr3automl.initial_design_runtime", initial_design_fraction = 0.5)