diff --git a/NAMESPACE b/NAMESPACE
index 53dbbfee4..53c328ef9 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -41,6 +41,7 @@ export(LearnerClassifRandomForest)
 export(LearnerClassifRandomForestSRC)
 export(LearnerClassifRandomForestWeka)
 export(LearnerClassifRandomPlantedForest)
+export(LearnerClassifTabNet)
 export(LearnerClassifRandomTree)
 export(LearnerClassifSGD)
 export(LearnerClassifSMO)
@@ -91,6 +92,7 @@ export(LearnerRegrRandomForest)
 export(LearnerRegrRandomForestSRC)
 export(LearnerRegrRandomForestWeka)
 export(LearnerRegrRandomPlantedForest)
+export(LearnerRegrTabNet)
 export(LearnerRegrRandomTree)
 export(LearnerRegrSGD)
 export(LearnerRegrSMOreg)
diff --git a/R/bibentries.R b/R/bibentries.R
index 77360bf3e..44305d780 100644
--- a/R/bibentries.R
+++ b/R/bibentries.R
@@ -586,6 +586,14 @@ bibentries = c( # nolint start
     month = "01",
     journal = "University of California, Berkeley"
   ),
+  arik2021tabnet = bibentry("inproceedings",
+    title = "Tabnet: Attentive interpretable tabular learning",
+    author = "Ar\xc4\xb1k, Sercan O and Pfister, Tomas",
+    booktitle = "AAAI",
+    volume = "35",
+    pages = "6679--6687",
+    year = "2021"
+  ),
   barnwal2022 = bibentry("article",
     title = "Survival Regression with Accelerated Failure Time Model in XGBoost",
     author = "Barnwal Avinash, Cho Hyunsu and Hocking Toby",
diff --git a/R/learner_tabnet_classif_tabnet.R b/R/learner_tabnet_classif_tabnet.R
new file mode 100644
index 000000000..0efbb6afb
--- /dev/null
+++ b/R/learner_tabnet_classif_tabnet.R
@@ -0,0 +1,102 @@
+#' @title Classification TabNet Learner
+#' @author Lukas Burk
+#' @name mlr_learners_classif.tabnet
+#'
+#' @template class_learner
+#' @templateVar id classif.tabnet
+#' @templateVar caller tabnet
+#' @references
+#' `r format_bib("arik2021tabnet")`
+#'
+#' @template seealso_learner
+#' @export
+#' @examples
+#' \dontrun{
+#' library(mlr3)
+#' library(mlr3torch)
+#' task = tsk("german_credit")
+#' lrn = lrn("classif.tabnet")
+#'
+#' lrn$param_set$values$epochs = 10
+#' lrn$param_set$values$attention_width = 8
+#' lrn$train(task)
+#' }
+LearnerClassifTabNet = R6Class("LearnerClassifTabNet",
+  inherit = LearnerClassif,
+  public = list(
+    #' @description
+    #' Creates a new instance of this [R6][R6::R6Class] class.
+    initialize = function() {
+      ps = params_tabnet()
+      super$initialize(
+        id = "classif.tabnet",
+        packages = "tabnet",
+        feature_types = c("logical", "integer", "numeric", "factor", "ordered"),
+        predict_types = c("response", "prob"),
+        param_set = ps,
+        properties = c("importance", "multiclass", "twoclass"),
+        man = "mlr3extralearners::mlr_learners_classif.tabnet",
+        label = "Attentive Interpretable Tabular Network"
+      )
+    },
+
+    #' @description
+    #' The importance scores are extracted from the slot `.$model$fit$importances`.
+    #' @return Named `numeric()`.
+    importance = function() {
+      if (is.null(self$model)) {
+        stopf("No model stored")
+      }
+      imp = self$model$fit$importances
+      sort(stats::setNames(imp$importance, imp$variables), decreasing = TRUE)
+    }
+  ),
+  private = list(
+    .train = function(task) {
+      # get parameters for training
+      pars = self$param_set$get_values(tags = "train")
+      pars_threads = pars$num_threads
+      pars$num_threads = NULL
+
+      # Set number of threads
+      torch::torch_set_num_threads(pars_threads)
+
+      # set column names to ensure consistency in fit and predict
+      self$state$feature_names = task$feature_names
+
+      # use the mlr3misc::invoke function (it's similar to do.call())
+      mlr3misc::invoke(tabnet::tabnet_fit,
+        x = task$data(cols = task$feature_names),
+        y = task$data(cols = task$target_names),
+        .args = pars
+      )
+    },
+
+    .predict = function(task) {
+      # get parameters with tag "predict"
+      pars = self$param_set$get_values(tags = "predict")
+
+      # get newdata and ensure same ordering in train and predict
+      newdata = task$data(cols = self$state$feature_names)
+
+      if (self$predict_type == "response") {
+        pred = mlr3misc::invoke(predict, self$model, new_data = newdata,
+          type = "class", .args = pars)
+
+        list(response = pred[[".pred_class"]])
+      } else {
+        pred = mlr3misc::invoke(predict, self$model, new_data = newdata,
+          type = "prob", .args = pars)
+
+        # Result will be a df with one column per variable with names '.pred_<level>'
+        # we want the names without ".pred"
+        names(pred) = sub(pattern = ".pred_", replacement = "", names(pred))
+
+        list(prob = as.matrix(pred))
+      }
+
+    }
+  )
+)
+
+.extralrns_dict$add("classif.tabnet", LearnerClassifTabNet)
diff --git a/R/learner_tabnet_regr_tabnet.R b/R/learner_tabnet_regr_tabnet.R
new file mode 100644
index 000000000..08d8491e7
--- /dev/null
+++ b/R/learner_tabnet_regr_tabnet.R
@@ -0,0 +1,97 @@
+#' @title Regression TabNet Learner
+#' @author Lukas Burk
+#' @name mlr_learners_regr.tabnet
+#'
+#' @template class_learner
+#' @templateVar id regr.tabnet
+#' @templateVar caller tabnet
+#' @references
+#' `r format_bib("arik2021tabnet")`
+#'
+#' @template seealso_learner
+#' @export
+#' @examples
+#' \dontrun{
+#' library(mlr3)
+#' library(mlr3torch)
+#'
+#' task = tsk("boston_housing")
+#'
+#' # Creating a learner & training on example task
+#' lrn = lrn("regr.tabnet")
+#'
+#' lrn$param_set$values$epochs = 10
+#' lrn$train(task)
+#'
+#' # Predict on training data, get RMSE
+#' predictions = lrn$predict(task)
+#' predictions$score(msr("regr.rmse"))
+#' }
+LearnerRegrTabNet = R6::R6Class("LearnerRegrTabnet",
+  inherit = LearnerRegr,
+  public = list(
+    #' @description
+    #' Creates a new instance of this [R6][R6::R6Class] class.
+    initialize = function() {
+      ps = params_tabnet()
+
+      super$initialize(
+        id = "regr.tabnet",
+        packages = "tabnet",
+        feature_types = c("logical", "integer", "numeric", "factor", "ordered"),
+        param_set = ps,
+        properties = c("importance"),
+        man = "mlr3torch::mlr_learners_regr.tabnet",
+        label = "Attentive Interpretable Tabular Network"
+      )
+    },
+
+    #' @description
+    #' The importance scores are extracted from the slot `.$model$fit$importances`.
+    #' @return Named `numeric()`.
+    importance = function() {
+      if (is.null(self$model)) {
+        stopf("No model stored")
+      }
+      imp = self$model$fit$importances
+      sort(stats::setNames(imp$importance, imp$variables), decreasing = TRUE)
+    }
+  ),
+  private = list(
+    .train = function(task) {
+      # get parameters for training
+      pars = self$param_set$get_values(tags = "train")
+      pars_threads = pars$num_threads
+      pars$num_threads = NULL
+
+      # Set number of threads
+      torch::torch_set_num_threads(pars_threads)
+
+      # set column names to ensure consistency in fit and predict
+      self$state$feature_names = task$feature_names
+
+      # use the mlr3misc::invoke function (it's similar to do.call())
+      mlr3misc::invoke(tabnet::tabnet_fit,
+                       x = task$data(cols = task$feature_names),
+                       y = task$data(cols = task$target_names),
+                       .args = pars
+      )
+    },
+    .predict = function(task) {
+      # get parameters with tag "predict"
+      pars = self$param_set$get_values(tags = "predict")
+
+      # get newdata and ensure same ordering in train and predict
+      newdata = task$data(cols = self$state$feature_names)
+
+      pred = mlr3misc::invoke(predict, self$model,
+                              new_data = newdata,
+                              .args = pars
+      )
+
+      list(response = pred[[".pred"]])
+    }
+  )
+)
+
+.extralrns_dict$add("regr.tabnet", LearnerRegrTabNet)
diff --git a/R/paramset_tabnet.R b/R/paramset_tabnet.R
new file mode 100644
index 000000000..ebcf71543
--- /dev/null
+++ b/R/paramset_tabnet.R
@@ -0,0 +1,48 @@
+params_tabnet = function() {
+  param_set = ParamSet$new(list(
+    num_threads= p_int(default = 1L, lower = 1L, upper = Inf, tags = c("train", "threads")),
+    batch_size = p_int(default = 256L, lower = 1L, upper = Inf, tags = "train"),
+    penalty = p_dbl(default = 0.001, tags = "train"),
+
+    # FIXME: NULL here is used for bool FALSE, not sure what to do there.
+    clip_value = p_uty(default = NULL, tags = "train"),
+    loss = p_fct(default = "auto", levels = c("auto", "mse", "cross_entropy"), tags = "train"),
+    epochs = p_int(default = 5L, lower = 1L, upper = Inf, tags = "train"),
+    drop_last = p_lgl(default = FALSE, tags = "train"),
+    decision_width = p_int(default = 8L, lower = 1L, upper = Inf, tags = "train"),
+    attention_width = p_int(default = 8L, lower = 1L, upper = Inf, tags = "train"),
+    num_steps = p_int(default = 3L, lower = 1L, upper = Inf, tags = "train"),
+    feature_reusage = p_dbl(default = 1.3, lower = 0, upper = Inf, tags = "train"),
+    mask_typs = p_fct(default = "sparsemax", levels = c("sparsemax", "entmax"), tags = "train"),
+    virtual_batch_size = p_int(default = 128L, lower = 1L, upper = Inf, tags = "train"),
+    valid_split = p_dbl(default = 0, lower = 0, upper = 1, tags = "train"),
+    learn_rate = p_dbl(default = 0.02, lower = 0, upper = 1, tags = "train"),
+
+    # FIXME: Currently either 'adam' or arbitrary optimizer function according to docs
+    optimizer = p_uty(default = "adam", tags = "train"),
+
+    # FIXME: This is either NULL or a function or explicit "steps", needs custom_check fun
+    lr_scheduler = p_uty(default = NULL, tags = "train"),
+
+    lr_decay = p_dbl(default = 0.1, lower = 0, upper = 1, tags = "train"),
+    step_size = p_int(default = 30L, lower = 1L, upper = Inf, tags = "train"),
+    checkpoint_epochs = p_int(default = 10L, lower = 1L, upper = Inf, tags = "train"),
+    cat_emb_dim = p_int(default = 1L, lower = 0L, upper = Inf, tags = "train"),
+    num_independent = p_int(default = 2L, lower = 0, upper = Inf, tags = "train"),
+    num_shared = p_int(default = 2L, lower = 0, upper = Inf, tags = "train"),
+    momentum = p_dbl(default = 0.02, lower = 0, upper = 1, tags = "train"),
+    pretraining_ratio = p_dbl(default = 0.5, lower = 0, upper = 1, tags = "train"),
+    verbose = p_lgl(default = FALSE, tags = "train"),
+    devics = p_fct(default = "auto", levels = c("auto", "cpu", "cuda"), tags = "train"),
+    importance_sample_size = p_int(lower = 0, upper = 1e5, special_vals = list(NULL), tags = "train")
+  ))
+
+  # Set param values that differ from default in tabnet_fit
+  param_set$values = list(
+    num_threads = 1L,
+    # clip_value = NULL,
+    decision_width = 8L,
+    attention_width = 8L
+  )
+  return(param_set)
+}
diff --git a/man/mlr_learners_classif.tabnet.Rd b/man/mlr_learners_classif.tabnet.Rd
new file mode 100644
index 000000000..cb435fe9a
--- /dev/null
+++ b/man/mlr_learners_classif.tabnet.Rd
@@ -0,0 +1,162 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/learner_tabnet_classif_tabnet.R
+\name{mlr_learners_classif.tabnet}
+\alias{mlr_learners_classif.tabnet}
+\alias{LearnerClassifTabNet}
+\title{Classification TabNet Learner}
+\description{
+Classification TabNet Learner
+
+Classification TabNet Learner
+}
+\section{Dictionary}{
+ This \link{Learner} can be instantiated via the
+\link[mlr3misc:Dictionary]{dictionary} \link{mlr_learners} or with the associated
+sugar function \code{\link[=lrn]{lrn()}}:
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{mlr_learners$get("classif.tabnet")
+lrn("classif.tabnet")
+}\if{html}{\out{</div>}}
+}
+
+\section{Meta Information}{
+
+\itemize{
+\item Task type: \dQuote{classif}
+\item Predict Types: \dQuote{response}, \dQuote{prob}
+\item Feature Types: \dQuote{logical}, \dQuote{integer}, \dQuote{numeric}, \dQuote{factor}, \dQuote{ordered}
+\item Required Packages: \CRANpkg{mlr3}, \CRANpkg{tabnet}
+}
+}
+
+\section{Parameters}{
+\tabular{lllll}{
+   Id \tab Type \tab Default \tab Levels \tab Range \cr
+   num_threads \tab integer \tab 1 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   batch_size \tab integer \tab 256 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   penalty \tab numeric \tab 0.001 \tab  \tab \eqn{(-\infty, \infty)}{(-Inf, Inf)} \cr
+   clip_value \tab untyped \tab  \tab  \tab - \cr
+   loss \tab character \tab auto \tab auto, mse, cross_entropy \tab - \cr
+   epochs \tab integer \tab 5 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   drop_last \tab logical \tab FALSE \tab TRUE, FALSE \tab - \cr
+   decision_width \tab integer \tab 8 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   attention_width \tab integer \tab 8 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   num_steps \tab integer \tab 3 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   feature_reusage \tab numeric \tab 1.3 \tab  \tab \eqn{[0, \infty)}{[0, Inf)} \cr
+   mask_type \tab character \tab sparsemax \tab sparsemax, entmax \tab - \cr
+   virtual_batch_size \tab integer \tab 128 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   valid_split \tab numeric \tab 0 \tab  \tab \eqn{[0, 1]}{[0, 1]} \cr
+   learn_rate \tab numeric \tab 0.02 \tab  \tab \eqn{[0, 1]}{[0, 1]} \cr
+   optimizer \tab untyped \tab adam \tab  \tab - \cr
+   lr_scheduler \tab untyped \tab  \tab  \tab - \cr
+   lr_decay \tab numeric \tab 0.1 \tab  \tab \eqn{[0, 1]}{[0, 1]} \cr
+   step_size \tab integer \tab 30 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   checkpoint_epochs \tab integer \tab 10 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   cat_emb_dim \tab integer \tab 1 \tab  \tab \eqn{[0, \infty)}{[0, Inf)} \cr
+   num_independent \tab integer \tab 2 \tab  \tab \eqn{[0, \infty)}{[0, Inf)} \cr
+   num_shared \tab integer \tab 2 \tab  \tab \eqn{[0, \infty)}{[0, Inf)} \cr
+   momentum \tab numeric \tab 0.02 \tab  \tab \eqn{[0, 1]}{[0, 1]} \cr
+   pretraining_ratio \tab numeric \tab 0.5 \tab  \tab \eqn{[0, 1]}{[0, 1]} \cr
+   verbose \tab logical \tab FALSE \tab TRUE, FALSE \tab - \cr
+   device \tab character \tab auto \tab auto, cpu, cuda \tab - \cr
+   importance_sample_size \tab integer \tab - \tab  \tab \eqn{[0, 1e+05]}{[0, 1e+05]} \cr
+}
+}
+
+\examples{
+\dontrun{
+library(mlr3)
+library(mlr3torch)
+task = tsk("german_credit")
+lrn = lrn("classif.tabnet")
+
+lrn$param_set$values$epochs = 10
+lrn$param_set$values$attention_width = 8
+lrn$train(task)
+}
+}
+\references{
+Arık, O S, Pfister, Tomas (2021).
+\dQuote{Tabnet: Attentive interpretable tabular learning.}
+In \emph{AAAI}, volume 35, 6679--6687.
+}
+\seealso{
+\itemize{
+\item \link[mlr3misc:Dictionary]{Dictionary} of \link[mlr3:Learner]{Learners}: \link[mlr3:mlr_learners]{mlr3::mlr_learners}.
+\item \code{as.data.table(mlr_learners)} for a table of available \link[=Learner]{Learners} in the running session (depending on the loaded packages).
+\item Chapter in the \href{https://mlr3book.mlr-org.com/}{mlr3book}: \url{https://mlr3book.mlr-org.com/basics.html#learners}
+\item \CRANpkg{mlr3learners} for a selection of recommended learners.
+\item \CRANpkg{mlr3cluster} for unsupervised clustering learners.
+\item \CRANpkg{mlr3pipelines} to combine learners with pre- and postprocessing steps.
+\item \CRANpkg{mlr3tuning} for tuning of hyperparameters, \CRANpkg{mlr3tuningspaces} for established default tuning spaces.
+}
+}
+\author{
+Lukas Burk
+}
+\section{Super classes}{
+\code{\link[mlr3:Learner]{mlr3::Learner}} -> \code{\link[mlr3:LearnerClassif]{mlr3::LearnerClassif}} -> \code{LearnerClassifTabNet}
+}
+\section{Methods}{
+\subsection{Public methods}{
+\itemize{
+\item \href{#method-LearnerClassifTabNet-new}{\code{LearnerClassifTabNet$new()}}
+\item \href{#method-LearnerClassifTabNet-importance}{\code{LearnerClassifTabNet$importance()}}
+\item \href{#method-LearnerClassifTabNet-clone}{\code{LearnerClassifTabNet$clone()}}
+}
+}
+\if{html}{\out{
+<details><summary>Inherited methods</summary>
+<ul>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="base_learner"><a href='../../mlr3/html/Learner.html#method-Learner-base_learner'><code>mlr3::Learner$base_learner()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="format"><a href='../../mlr3/html/Learner.html#method-Learner-format'><code>mlr3::Learner$format()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="help"><a href='../../mlr3/html/Learner.html#method-Learner-help'><code>mlr3::Learner$help()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="predict"><a href='../../mlr3/html/Learner.html#method-Learner-predict'><code>mlr3::Learner$predict()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="predict_newdata"><a href='../../mlr3/html/Learner.html#method-Learner-predict_newdata'><code>mlr3::Learner$predict_newdata()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="print"><a href='../../mlr3/html/Learner.html#method-Learner-print'><code>mlr3::Learner$print()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="reset"><a href='../../mlr3/html/Learner.html#method-Learner-reset'><code>mlr3::Learner$reset()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="train"><a href='../../mlr3/html/Learner.html#method-Learner-train'><code>mlr3::Learner$train()</code></a></span></li>
+</ul>
+</details>
+}}
+\if{html}{\out{<hr>}}
+\if{html}{\out{<a id="method-LearnerClassifTabNet-new"></a>}}
+\if{latex}{\out{\hypertarget{method-LearnerClassifTabNet-new}{}}}
+\subsection{Method \code{new()}}{
+Creates a new instance of this \link[R6:R6Class]{R6} class.
+\subsection{Usage}{
+\if{html}{\out{<div class="r">}}\preformatted{LearnerClassifTabNet$new()}\if{html}{\out{</div>}}
+}
+
+}
+\if{html}{\out{<hr>}}
+\if{html}{\out{<a id="method-LearnerClassifTabNet-importance"></a>}}
+\if{latex}{\out{\hypertarget{method-LearnerClassifTabNet-importance}{}}}
+\subsection{Method \code{importance()}}{
+The importance scores are extracted from the slot \code{.$model$fit$importances}.
+\subsection{Usage}{
+\if{html}{\out{<div class="r">}}\preformatted{LearnerClassifTabNet$importance()}\if{html}{\out{</div>}}
+}
+
+\subsection{Returns}{
+Named \code{numeric()}.
+}
+}
+\if{html}{\out{<hr>}}
+\if{html}{\out{<a id="method-LearnerClassifTabNet-clone"></a>}}
+\if{latex}{\out{\hypertarget{method-LearnerClassifTabNet-clone}{}}}
+\subsection{Method \code{clone()}}{
+The objects of this class are cloneable with this method.
+\subsection{Usage}{
+\if{html}{\out{<div class="r">}}\preformatted{LearnerClassifTabNet$clone(deep = FALSE)}\if{html}{\out{</div>}}
+}
+
+\subsection{Arguments}{
+\if{html}{\out{<div class="arguments">}}
+\describe{
+\item{\code{deep}}{Whether to make a deep clone.}
+}
+\if{html}{\out{</div>}}
+}
+}
+}
diff --git a/man/mlr_learners_regr.tabnet.Rd b/man/mlr_learners_regr.tabnet.Rd
new file mode 100644
index 000000000..c247c7eb6
--- /dev/null
+++ b/man/mlr_learners_regr.tabnet.Rd
@@ -0,0 +1,168 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/learner_tabnet_regr_tabnet.R
+\name{mlr_learners_regr.tabnet}
+\alias{mlr_learners_regr.tabnet}
+\alias{LearnerRegrTabNet}
+\title{Regression TabNet Learner}
+\description{
+Regression TabNet Learner
+
+Regression TabNet Learner
+}
+\section{Dictionary}{
+ This \link{Learner} can be instantiated via the
+\link[mlr3misc:Dictionary]{dictionary} \link{mlr_learners} or with the associated
+sugar function \code{\link[=lrn]{lrn()}}:
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{mlr_learners$get("regr.tabnet")
+lrn("regr.tabnet")
+}\if{html}{\out{</div>}}
+}
+
+\section{Meta Information}{
+
+\itemize{
+\item Task type: \dQuote{regr}
+\item Predict Types: \dQuote{response}
+\item Feature Types: \dQuote{logical}, \dQuote{integer}, \dQuote{numeric}, \dQuote{factor}, \dQuote{ordered}
+\item Required Packages: \CRANpkg{mlr3}, \CRANpkg{tabnet}
+}
+}
+
+\section{Parameters}{
+\tabular{lllll}{
+   Id \tab Type \tab Default \tab Levels \tab Range \cr
+   num_threads \tab integer \tab 1 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   batch_size \tab integer \tab 256 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   penalty \tab numeric \tab 0.001 \tab  \tab \eqn{(-\infty, \infty)}{(-Inf, Inf)} \cr
+   clip_value \tab untyped \tab  \tab  \tab - \cr
+   loss \tab character \tab auto \tab auto, mse, cross_entropy \tab - \cr
+   epochs \tab integer \tab 5 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   drop_last \tab logical \tab FALSE \tab TRUE, FALSE \tab - \cr
+   decision_width \tab integer \tab 8 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   attention_width \tab integer \tab 8 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   num_steps \tab integer \tab 3 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   feature_reusage \tab numeric \tab 1.3 \tab  \tab \eqn{[0, \infty)}{[0, Inf)} \cr
+   mask_type \tab character \tab sparsemax \tab sparsemax, entmax \tab - \cr
+   virtual_batch_size \tab integer \tab 128 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   valid_split \tab numeric \tab 0 \tab  \tab \eqn{[0, 1]}{[0, 1]} \cr
+   learn_rate \tab numeric \tab 0.02 \tab  \tab \eqn{[0, 1]}{[0, 1]} \cr
+   optimizer \tab untyped \tab adam \tab  \tab - \cr
+   lr_scheduler \tab untyped \tab  \tab  \tab - \cr
+   lr_decay \tab numeric \tab 0.1 \tab  \tab \eqn{[0, 1]}{[0, 1]} \cr
+   step_size \tab integer \tab 30 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   checkpoint_epochs \tab integer \tab 10 \tab  \tab \eqn{[1, \infty)}{[1, Inf)} \cr
+   cat_emb_dim \tab integer \tab 1 \tab  \tab \eqn{[0, \infty)}{[0, Inf)} \cr
+   num_independent \tab integer \tab 2 \tab  \tab \eqn{[0, \infty)}{[0, Inf)} \cr
+   num_shared \tab integer \tab 2 \tab  \tab \eqn{[0, \infty)}{[0, Inf)} \cr
+   momentum \tab numeric \tab 0.02 \tab  \tab \eqn{[0, 1]}{[0, 1]} \cr
+   pretraining_ratio \tab numeric \tab 0.5 \tab  \tab \eqn{[0, 1]}{[0, 1]} \cr
+   verbose \tab logical \tab FALSE \tab TRUE, FALSE \tab - \cr
+   device \tab character \tab auto \tab auto, cpu, cuda \tab - \cr
+   importance_sample_size \tab integer \tab - \tab  \tab \eqn{[0, 1e+05]}{[0, 1e+05]} \cr
+}
+}
+
+\examples{
+\dontrun{
+library(mlr3)
+library(mlr3torch)
+
+task = tsk("boston_housing")
+
+# Creating a learner & training on example task
+lrn = lrn("regr.tabnet")
+
+lrn$param_set$values$epochs = 10
+lrn$train(task)
+
+# Predict on training data, get RMSE
+predictions = lrn$predict(task)
+predictions$score(msr("regr.rmse"))
+}
+}
+\references{
+Arık, O S, Pfister, Tomas (2021).
+\dQuote{Tabnet: Attentive interpretable tabular learning.}
+In \emph{AAAI}, volume 35, 6679--6687.
+}
+\seealso{
+\itemize{
+\item \link[mlr3misc:Dictionary]{Dictionary} of \link[mlr3:Learner]{Learners}: \link[mlr3:mlr_learners]{mlr3::mlr_learners}.
+\item \code{as.data.table(mlr_learners)} for a table of available \link[=Learner]{Learners} in the running session (depending on the loaded packages).
+\item Chapter in the \href{https://mlr3book.mlr-org.com/}{mlr3book}: \url{https://mlr3book.mlr-org.com/basics.html#learners}
+\item \CRANpkg{mlr3learners} for a selection of recommended learners.
+\item \CRANpkg{mlr3cluster} for unsupervised clustering learners.
+\item \CRANpkg{mlr3pipelines} to combine learners with pre- and postprocessing steps.
+\item \CRANpkg{mlr3tuning} for tuning of hyperparameters, \CRANpkg{mlr3tuningspaces} for established default tuning spaces.
+}
+}
+\author{
+Lukas Burk
+}
+\section{Super classes}{
+\code{\link[mlr3:Learner]{mlr3::Learner}} -> \code{\link[mlr3:LearnerRegr]{mlr3::LearnerRegr}} -> \code{LearnerRegrTabnet}
+}
+\section{Methods}{
+\subsection{Public methods}{
+\itemize{
+\item \href{#method-LearnerRegrTabnet-new}{\code{LearnerRegrTabNet$new()}}
+\item \href{#method-LearnerRegrTabnet-importance}{\code{LearnerRegrTabNet$importance()}}
+\item \href{#method-LearnerRegrTabnet-clone}{\code{LearnerRegrTabNet$clone()}}
+}
+}
+\if{html}{\out{
+<details><summary>Inherited methods</summary>
+<ul>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="base_learner"><a href='../../mlr3/html/Learner.html#method-Learner-base_learner'><code>mlr3::Learner$base_learner()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="format"><a href='../../mlr3/html/Learner.html#method-Learner-format'><code>mlr3::Learner$format()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="help"><a href='../../mlr3/html/Learner.html#method-Learner-help'><code>mlr3::Learner$help()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="predict"><a href='../../mlr3/html/Learner.html#method-Learner-predict'><code>mlr3::Learner$predict()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="predict_newdata"><a href='../../mlr3/html/Learner.html#method-Learner-predict_newdata'><code>mlr3::Learner$predict_newdata()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="print"><a href='../../mlr3/html/Learner.html#method-Learner-print'><code>mlr3::Learner$print()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="reset"><a href='../../mlr3/html/Learner.html#method-Learner-reset'><code>mlr3::Learner$reset()</code></a></span></li>
+<li><span class="pkg-link" data-pkg="mlr3" data-topic="Learner" data-id="train"><a href='../../mlr3/html/Learner.html#method-Learner-train'><code>mlr3::Learner$train()</code></a></span></li>
+</ul>
+</details>
+}}
+\if{html}{\out{<hr>}}
+\if{html}{\out{<a id="method-LearnerRegrTabnet-new"></a>}}
+\if{latex}{\out{\hypertarget{method-LearnerRegrTabnet-new}{}}}
+\subsection{Method \code{new()}}{
+Creates a new instance of this \link[R6:R6Class]{R6} class.
+\subsection{Usage}{
+\if{html}{\out{<div class="r">}}\preformatted{LearnerRegrTabNet$new()}\if{html}{\out{</div>}}
+}
+
+}
+\if{html}{\out{<hr>}}
+\if{html}{\out{<a id="method-LearnerRegrTabnet-importance"></a>}}
+\if{latex}{\out{\hypertarget{method-LearnerRegrTabnet-importance}{}}}
+\subsection{Method \code{importance()}}{
+The importance scores are extracted from the slot \code{.$model$fit$importances}.
+\subsection{Usage}{
+\if{html}{\out{<div class="r">}}\preformatted{LearnerRegrTabNet$importance()}\if{html}{\out{</div>}}
+}
+
+\subsection{Returns}{
+Named \code{numeric()}.
+}
+}
+\if{html}{\out{<hr>}}
+\if{html}{\out{<a id="method-LearnerRegrTabnet-clone"></a>}}
+\if{latex}{\out{\hypertarget{method-LearnerRegrTabnet-clone}{}}}
+\subsection{Method \code{clone()}}{
+The objects of this class are cloneable with this method.
+\subsection{Usage}{
+\if{html}{\out{<div class="r">}}\preformatted{LearnerRegrTabNet$clone(deep = FALSE)}\if{html}{\out{</div>}}
+}
+
+\subsection{Arguments}{
+\if{html}{\out{<div class="arguments">}}
+\describe{
+\item{\code{deep}}{Whether to make a deep clone.}
+}
+\if{html}{\out{</div>}}
+}
+}
+}
diff --git a/man/params_tabnet.Rd b/man/params_tabnet.Rd
new file mode 100644
index 000000000..88c35669f
--- /dev/null
+++ b/man/params_tabnet.Rd
@@ -0,0 +1,15 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/paramset_tabnet.R
+\name{params_tabnet}
+\alias{params_tabnet}
+\title{Wrapper for the TabNet ParamSet}
+\usage{
+params_tabnet()
+}
+\value{
+Object of classes \verb{ParamSet" "R6"}, suitable for use with
+}
+\description{
+Used to de-duplicate learner setup.
+}
+\keyword{internal}
diff --git a/tests/testthat/helper.R b/tests/testthat/helper.R
index 54d6785a2..ee21de7b1 100644
--- a/tests/testthat/helper.R
+++ b/tests/testthat/helper.R
@@ -3,8 +3,10 @@ library(mlr3)
 library(mlr3extralearners)
 library(mlr3proba)
 
-lapply(list.files(system.file("testthat", package = "mlr3"),
-  pattern = "^helper.*\\.[rR]", full.names = TRUE), source)
+tmp = list.files(system.file("testthat", package = "mlr3"),
+  pattern = "^helper.*\\.[rR]", full.names = TRUE)
+tmp = tmp[basename(tmp) != "helper_debugging.R"]
+lapply(tmp, source)
 
 lapply(list.files(system.file("testthat", package = "mlr3proba"),
   pattern = "^helper.*\\.[rR]", full.names = TRUE), source)
diff --git a/tests/testthat/test_paramset_tabnet_regr_tabnet.R b/tests/testthat/test_paramset_tabnet_regr_tabnet.R
new file mode 100644
index 000000000..c949b07c5
--- /dev/null
+++ b/tests/testthat/test_paramset_tabnet_regr_tabnet.R
@@ -0,0 +1,21 @@
+test_that("paramtest regr.tabnet train", {
+  learner = lrn("regr.tabnet")
+  fun = tabnet::tabnet_fit # nolint
+
+  exclude = c(
+  )
+
+  paramtest = run_paramtest(learner, fun, exclude, tag = "train")
+  expect_paramtest(paramtest)
+})
+
+test_that("paramtest regr.earth predict", {
+  learner = lrn("regr.tabnet")
+  fun = tabnet:::predict.tabnet_fit # nolint
+
+  exclude = c(
+  )
+
+  paramtest = run_paramtest(learner, fun, exclude, tag = "predict")
+  expect_paramtest(paramtest)
+})
diff --git a/tests/testthat/test_tabnet_classif_tabnet.R b/tests/testthat/test_tabnet_classif_tabnet.R
new file mode 100644
index 000000000..8381951c1
--- /dev/null
+++ b/tests/testthat/test_tabnet_classif_tabnet.R
@@ -0,0 +1,6 @@
+test_that("autotest", {
+  learner = lrn("classif.tabnet", device = "cpu", epochs = 20)
+  expect_learner(learner)
+  result = run_autotest(learner)
+  expect_true(result, info = result$error)
+})
diff --git a/tests/testthat/test_tabnet_regr_tabnet.R b/tests/testthat/test_tabnet_regr_tabnet.R
new file mode 100644
index 000000000..fa5f75c64
--- /dev/null
+++ b/tests/testthat/test_tabnet_regr_tabnet.R
@@ -0,0 +1,6 @@
+test_that("autotest", {
+  learner = lrn("regr.tabnet", device = "cpu", epochs = 30)
+  expect_learner(learner)
+  result = run_autotest(learner)
+  expect_true(result, info = result$error)
+})