diff --git a/NAMESPACE b/NAMESPACE
index d6b6b37f..973e19dc 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -11,12 +11,14 @@ S3method(handler_predict,default)
 S3method(handler_predict,glm)
 S3method(handler_predict,lm)
 S3method(handler_predict,ranger)
+S3method(handler_predict,tabnet_fit)
 S3method(handler_predict,train)
 S3method(handler_predict,workflow)
 S3method(handler_predict,xgb.Booster)
 S3method(handler_startup,Learner)
 S3method(handler_startup,default)
 S3method(handler_startup,ranger)
+S3method(handler_startup,tabnet_fit)
 S3method(handler_startup,train)
 S3method(handler_startup,workflow)
 S3method(handler_startup,xgb.Booster)
@@ -32,12 +34,14 @@ S3method(vetiver_create_description,default)
 S3method(vetiver_create_description,glm)
 S3method(vetiver_create_description,lm)
 S3method(vetiver_create_description,ranger)
+S3method(vetiver_create_description,tabnet_fit)
 S3method(vetiver_create_description,train)
 S3method(vetiver_create_description,workflow)
 S3method(vetiver_create_description,xgb.Booster)
 S3method(vetiver_create_meta,Learner)
 S3method(vetiver_create_meta,default)
 S3method(vetiver_create_meta,ranger)
+S3method(vetiver_create_meta,tabnet_fit)
 S3method(vetiver_create_meta,train)
 S3method(vetiver_create_meta,workflow)
 S3method(vetiver_create_meta,xgb.Booster)
@@ -46,6 +50,7 @@ S3method(vetiver_prepare_model,default)
 S3method(vetiver_prepare_model,glm)
 S3method(vetiver_prepare_model,lm)
 S3method(vetiver_prepare_model,ranger)
+S3method(vetiver_prepare_model,tabnet_fit)
 S3method(vetiver_prepare_model,train)
 S3method(vetiver_prepare_model,workflow)
 S3method(vetiver_ptype,Learner)
@@ -53,6 +58,7 @@ S3method(vetiver_ptype,default)
 S3method(vetiver_ptype,glm)
 S3method(vetiver_ptype,lm)
 S3method(vetiver_ptype,ranger)
+S3method(vetiver_ptype,tabnet_fit)
 S3method(vetiver_ptype,train)
 S3method(vetiver_ptype,workflow)
 S3method(vetiver_ptype,xgb.Booster)
diff --git a/NEWS.md b/NEWS.md
index 7e1b4a73..3a214578 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -1,5 +1,7 @@
 # vetiver (development version)
 
+* Added support for tabnet models (#124 @cregouby)
+
 * Trailing slashes are now removed from `vetiver_endpoint()` (#134).
 
 # vetiver 0.1.7
diff --git a/R/tabnet.R b/R/tabnet.R
new file mode 100644
index 00000000..5f398d78
--- /dev/null
+++ b/R/tabnet.R
@@ -0,0 +1,54 @@
+#' @rdname vetiver_create_description
+#' @export
+vetiver_create_description.tabnet_fit <- function(model) {
+    paste0(
+        "A tabnet `nn_module` containing ",
+        format(sum(sapply(model$fit$network$parameters, function(x) prod(x$shape))), nsmall = 0, big.mark = ",", scientific = FALSE),
+        " parameters."
+  )
+}
+
+#' @rdname vetiver_create_meta
+#' @export
+vetiver_create_meta.tabnet_fit <- function(model, metadata) {
+    vetiver_meta(metadata, required_pkgs = "tabnet")
+}
+
+#' @rdname vetiver_create_description
+#' @export
+vetiver_prepare_model.tabnet_fit <- function(model) {
+    butcher::butcher(model)
+}
+
+#' @rdname vetiver_create_ptype
+#' @export
+vetiver_ptype.tabnet_fit <- function(model, ...) {
+    rlang::check_dots_used()
+    dots <- list(...)
+    check_ptype_data(dots)
+    ptype <- vctrs::vec_ptype(dots$ptype_data)
+    tibble::as_tibble(ptype)
+}
+
+#' @rdname handler_startup
+#' @export
+handler_startup.tabnet_fit <- function(vetiver_model) {
+    attach_pkgs("tabnet")
+}
+
+#' @rdname handler_startup
+#' @export
+handler_predict.tabnet_fit <- function(vetiver_model, ...) {
+
+    ptype <- vetiver_model$blueprint$ptypes
+
+    function(req) {
+        new_data <- req$body
+        if (!is_null(ptype)) {
+            new_data <- vetiver_type_convert(new_data, ptype)
+            new_data <- hardhat::scream(new_data, ptype)
+        }
+        ret <- predict(vetiver_model, data = new_data, ...)
+        list(.pred = ret$.pred)
+    }
+}
diff --git a/man/handler_startup.Rd b/man/handler_startup.Rd
index 64ba6b9d..83bd98e3 100644
--- a/man/handler_startup.Rd
+++ b/man/handler_startup.Rd
@@ -1,6 +1,6 @@
 % Generated by roxygen2: do not edit by hand
 % Please edit documentation in R/caret.R, R/glm.R, R/handlers.R, R/lm.R,
-%   R/mlr3.R, R/ranger.R, R/tidymodels.R, R/xgboost.R
+%   R/mlr3.R, R/ranger.R, R/tabnet.R, R/tidymodels.R, R/xgboost.R
 \name{handler_startup.train}
 \alias{handler_startup.train}
 \alias{handler_predict.train}
@@ -14,6 +14,8 @@
 \alias{handler_predict.Learner}
 \alias{handler_startup.ranger}
 \alias{handler_predict.ranger}
+\alias{handler_startup.tabnet_fit}
+\alias{handler_predict.tabnet_fit}
 \alias{handler_startup.workflow}
 \alias{handler_predict.workflow}
 \alias{handler_startup.xgb.Booster}
@@ -44,6 +46,10 @@ handler_predict(vetiver_model, ...)
 
 \method{handler_predict}{ranger}(vetiver_model, ...)
 
+\method{handler_startup}{tabnet_fit}(vetiver_model)
+
+\method{handler_predict}{tabnet_fit}(vetiver_model, ...)
+
 \method{handler_startup}{workflow}(vetiver_model)
 
 \method{handler_predict}{workflow}(vetiver_model, ...)
diff --git a/man/vetiver-package.Rd b/man/vetiver-package.Rd
index 030ae4d6..f4314bb7 100644
--- a/man/vetiver-package.Rd
+++ b/man/vetiver-package.Rd
@@ -6,7 +6,7 @@
 \alias{vetiver-package}
 \title{vetiver: Version, Share, Deploy, and Monitor Models}
 \description{
-\if{html}{\figure{logo.png}{options: style='float: right' alt='logo' width='120'}}
+\if{html}{\figure{logo.png}{options: align='right' alt='logo' width='120'}}
 
 The goal of 'vetiver' is to provide fluent tooling to version, share, deploy, and monitor a trained model. Functions handle both recording and checking the model's input data prototype, and predicting from a remote API endpoint. The 'vetiver' package is extensible, with generics that can support many kinds of models.
 }
diff --git a/man/vetiver_compute_metrics.Rd b/man/vetiver_compute_metrics.Rd
index dd07eda6..1cb7926d 100644
--- a/man/vetiver_compute_metrics.Rd
+++ b/man/vetiver_compute_metrics.Rd
@@ -71,7 +71,15 @@ epoch time of \verb{1970-01-01 00:00:00}, \emph{in the time zone of the index}.
 This is generally used to define the anchor time to count from, which is
 relevant when the every value is \verb{> 1}.}
 
-\item{before, after}{\verb{[integer(1) / Inf]}
+\item{before}{\verb{[integer(1) / Inf]}
+
+The number of values before or after the current element to
+include in the sliding window. Set to \code{Inf} to select all elements
+before or after the current element. Negative values are allowed, which
+allows you to "look forward" from the current element if used as the
+\code{.before} value, or "look backwards" if used as \code{.after}.}
+
+\item{after}{\verb{[integer(1) / Inf]}
 
 The number of values before or after the current element to
 include in the sliding window. Set to \code{Inf} to select all elements
diff --git a/man/vetiver_create_description.Rd b/man/vetiver_create_description.Rd
index 051a8566..17af36de 100644
--- a/man/vetiver_create_description.Rd
+++ b/man/vetiver_create_description.Rd
@@ -1,6 +1,6 @@
 % Generated by roxygen2: do not edit by hand
 % Please edit documentation in R/caret.R, R/glm.R, R/lm.R, R/mlr3.R,
-%   R/prepare.R, R/ranger.R, R/tidymodels.R, R/xgboost.R
+%   R/prepare.R, R/ranger.R, R/tabnet.R, R/tidymodels.R, R/xgboost.R
 \name{vetiver_create_description.train}
 \alias{vetiver_create_description.train}
 \alias{vetiver_prepare_model.train}
@@ -16,6 +16,8 @@
 \alias{vetiver_prepare_model.default}
 \alias{vetiver_create_description.ranger}
 \alias{vetiver_prepare_model.ranger}
+\alias{vetiver_create_description.tabnet_fit}
+\alias{vetiver_prepare_model.tabnet_fit}
 \alias{vetiver_create_description.workflow}
 \alias{vetiver_prepare_model.workflow}
 \alias{vetiver_create_description.xgb.Booster}
@@ -49,6 +51,10 @@ vetiver_prepare_model(model)
 
 \method{vetiver_prepare_model}{ranger}(model)
 
+\method{vetiver_create_description}{tabnet_fit}(model)
+
+\method{vetiver_prepare_model}{tabnet_fit}(model)
+
 \method{vetiver_create_description}{workflow}(model)
 
 \method{vetiver_prepare_model}{workflow}(model)
diff --git a/man/vetiver_create_meta.Rd b/man/vetiver_create_meta.Rd
index 7af45fa6..43b1299a 100644
--- a/man/vetiver_create_meta.Rd
+++ b/man/vetiver_create_meta.Rd
@@ -1,6 +1,6 @@
 % Generated by roxygen2: do not edit by hand
 % Please edit documentation in R/caret.R, R/meta.R, R/mlr3.R, R/ranger.R,
-%   R/tidymodels.R, R/xgboost.R
+%   R/tabnet.R, R/tidymodels.R, R/xgboost.R
 \name{vetiver_create_meta.train}
 \alias{vetiver_create_meta.train}
 \alias{vetiver_meta}
@@ -8,6 +8,7 @@
 \alias{vetiver_create_meta.default}
 \alias{vetiver_create_meta.Learner}
 \alias{vetiver_create_meta.ranger}
+\alias{vetiver_create_meta.tabnet_fit}
 \alias{vetiver_create_meta.workflow}
 \alias{vetiver_create_meta.xgb.Booster}
 \title{Metadata constructors for \code{vetiver_model()} object}
@@ -24,6 +25,8 @@ vetiver_create_meta(model, metadata)
 
 \method{vetiver_create_meta}{ranger}(model, metadata)
 
+\method{vetiver_create_meta}{tabnet_fit}(model, metadata)
+
 \method{vetiver_create_meta}{workflow}(model, metadata)
 
 \method{vetiver_create_meta}{xgb.Booster}(model, metadata)
diff --git a/man/vetiver_create_ptype.Rd b/man/vetiver_create_ptype.Rd
index f1f43191..51dd2326 100644
--- a/man/vetiver_create_ptype.Rd
+++ b/man/vetiver_create_ptype.Rd
@@ -1,6 +1,6 @@
 % Generated by roxygen2: do not edit by hand
 % Please edit documentation in R/caret.R, R/glm.R, R/lm.R, R/mlr3.R, R/ptype.R,
-%   R/ranger.R, R/tidymodels.R, R/xgboost.R
+%   R/ranger.R, R/tabnet.R, R/tidymodels.R, R/xgboost.R
 \name{vetiver_ptype.train}
 \alias{vetiver_ptype.train}
 \alias{vetiver_ptype.glm}
@@ -10,6 +10,7 @@
 \alias{vetiver_ptype.default}
 \alias{vetiver_create_ptype}
 \alias{vetiver_ptype.ranger}
+\alias{vetiver_ptype.tabnet_fit}
 \alias{vetiver_ptype.workflow}
 \alias{vetiver_ptype.xgb.Booster}
 \title{Create a vetiver input data prototype}
@@ -30,6 +31,8 @@ vetiver_create_ptype(model, save_ptype, ...)
 
 \method{vetiver_ptype}{ranger}(model, ...)
 
+\method{vetiver_ptype}{tabnet_fit}(model, ...)
+
 \method{vetiver_ptype}{workflow}(model, ...)
 
 \method{vetiver_ptype}{xgb.Booster}(model, ...)
@@ -80,9 +83,4 @@ vetiver_ptype(cars_lm)
 
 ## can also turn off `ptype`
 vetiver_create_ptype(cars_lm, FALSE)
-\dontshow{if (rlang::is_installed("ranger")) (if (getRversion() >= "3.4") withAutoprint else force)(\{ # examplesIf}
-## some models require that you pass in training features
-cars_rf <- ranger::ranger(mpg ~ ., data = mtcars)
-vetiver_ptype(cars_rf, ptype_data = mtcars[,-1])
-\dontshow{\}) # examplesIf}
 }
diff --git a/man/vetiver_create_rsconnect_bundle.Rd b/man/vetiver_create_rsconnect_bundle.Rd
index be440534..acd107d8 100644
--- a/man/vetiver_create_rsconnect_bundle.Rd
+++ b/man/vetiver_create_rsconnect_bundle.Rd
@@ -46,23 +46,6 @@ The two functions \code{vetiver_create_rsconnect_bundle()} and
 \code{\link[=vetiver_deploy_rsconnect]{vetiver_deploy_rsconnect()}} are alternatives to each other, providing
 different strategies for deploying a vetiver model API to RStudio Connect.
 }
-\examples{
-\dontshow{if (rlang::is_installed("connectapi")) (if (getRversion() >= "3.4") withAutoprint else force)(\{ # examplesIf}
-library(pins)
-b <- board_temp(versioned = TRUE)
-cars_lm <- lm(mpg ~ ., data = mtcars)
-v <- vetiver_model(cars_lm, "cars_linear")
-vetiver_pin_write(b, v)
-
-## when you pin to RStudio Connect, your pin name will be typically be like:
-## "user.name/cars_linear"
-vetiver_create_rsconnect_bundle(
-    b,
-    "cars_linear",
-    predict_args = list(debug = TRUE)
-)
-\dontshow{\}) # examplesIf}
-}
 \seealso{
 \code{\link[=vetiver_write_plumber]{vetiver_write_plumber()}}, \code{\link[=vetiver_deploy_rsconnect]{vetiver_deploy_rsconnect()}}
 }
diff --git a/man/vetiver_dashboard.Rd b/man/vetiver_dashboard.Rd
index 7f66a962..8dcf25e8 100644
--- a/man/vetiver_dashboard.Rd
+++ b/man/vetiver_dashboard.Rd
@@ -41,7 +41,7 @@ helper function \code{pin_example_kc_housing_model()} to set up demonstration
 model and metrics pins needed for the monitoring demo. This function will:
 \itemize{
 \item fit an example model to training data
-\item pin the vetiver model to your own \code{\link[pins:board_folder]{pins::board_local()}}
+\item pin the vetiver model to your own \code{\link[pins:board_local]{pins::board_local()}}
 \item compute metrics from testing data
 \item pin these metrics to the same local board
 }
diff --git a/man/vetiver_deploy_rsconnect.Rd b/man/vetiver_deploy_rsconnect.Rd
index 457d04d1..110bd2ef 100644
--- a/man/vetiver_deploy_rsconnect.Rd
+++ b/man/vetiver_deploy_rsconnect.Rd
@@ -44,31 +44,6 @@ The two functions \code{vetiver_deploy_rsconnect()} and
 \code{\link[=vetiver_create_rsconnect_bundle]{vetiver_create_rsconnect_bundle()}} are alternatives to each other, providing
 different strategies for deploying a vetiver model API to RStudio Connect.
 }
-\examples{
-\dontshow{if (rlang::is_installed("rsconnect")) (if (getRversion() >= "3.4") withAutoprint else force)(\{ # examplesIf}
-library(pins)
-b <- board_temp(versioned = TRUE)
-cars_lm <- lm(mpg ~ ., data = mtcars)
-v <- vetiver_model(cars_lm, "cars_linear")
-vetiver_pin_write(b, v)
-
-if (FALSE) {
-## pass args for predicting:
-vetiver_deploy_rsconnect(
-    b,
-    "user.name/cars_linear",
-    predict_args = list(debug = TRUE)
-)
-
-## specify an account name through `...`:
-vetiver_deploy_rsconnect(
-    b,
-    "user.name/cars_linear",
-    account = "user.name"
-)
-}
-\dontshow{\}) # examplesIf}
-}
 \seealso{
 \code{\link[=vetiver_write_plumber]{vetiver_write_plumber()}}, \code{\link[=vetiver_create_rsconnect_bundle]{vetiver_create_rsconnect_bundle()}}
 }
diff --git a/man/vetiver_pr_predict.Rd b/man/vetiver_pr_predict.Rd
index 071c7b9b..532799b2 100644
--- a/man/vetiver_pr_predict.Rd
+++ b/man/vetiver_pr_predict.Rd
@@ -24,7 +24,7 @@ vetiver_pr_predict(
 \item{...}{Other arguments passed to \code{predict()}, such as prediction \code{type}}
 }
 \description{
-\ifelse{html}{\href{https://lifecycle.r-lib.org/articles/stages.html#deprecated}{\figure{lifecycle-deprecated.svg}{options: alt='[Deprecated]'}}}{\strong{[Deprecated]}}
+\verb{r lifecycle::badge("deprecated")}
 
 This function was deprecated to use \link{vetiver_api} directly instead.
 }
diff --git a/man/vetiver_write_docker.Rd b/man/vetiver_write_docker.Rd
index ff5b9ade..e792ddc4 100644
--- a/man/vetiver_write_docker.Rd
+++ b/man/vetiver_write_docker.Rd
@@ -45,22 +45,3 @@ After creating a Plumber file with \code{\link[=vetiver_write_plumber]{vetiver_w
 \code{vetiver_write_docker()} to create a Dockerfile plus a \code{vetiver_renv.lock}
 file for a pinned \code{\link[=vetiver_model]{vetiver_model()}}.
 }
-\examples{
-\dontshow{if (interactive() || identical(Sys.getenv("IN_PKGDOWN"), "true")) (if (getRversion() >= "3.4") withAutoprint else force)(\{ # examplesIf}
-
-library(pins)
-tmp_plumber <- tempfile()
-b <- board_temp(versioned = TRUE)
-cars_lm <- lm(mpg ~ ., data = mtcars)
-v <- vetiver_model(cars_lm, "cars_linear")
-vetiver_pin_write(b, v)
-vetiver_write_plumber(b, "cars_linear", file = tmp_plumber)
-
-## default port
-vetiver_write_docker(v, tmp_plumber, tempdir())
-## port from env variable
-vetiver_write_docker(v, tmp_plumber, tempdir(),
-                     port = 'as.numeric(Sys.getenv("PORT"))',
-                     expose = FALSE)
-\dontshow{\}) # examplesIf}
-}
diff --git a/tests/testthat/_snaps/tabnet.md b/tests/testthat/_snaps/tabnet.md
new file mode 100644
index 00000000..6933bc3d
--- /dev/null
+++ b/tests/testthat/_snaps/tabnet.md
@@ -0,0 +1,41 @@
+# can print tabnet model
+
+    Code
+      v
+    Output
+      
+      -- cars3 - <tabnet_fit> model for deployment 
+      A tabnet `nn_module` containing 6,301 parameters. using 10 features
+
+# error for no ptype_data with tabnet
+
+    Code
+      vetiver_model(car_tn, "cars3")
+    Condition
+      Error in `vetiver_create_description()`:
+      ! object 'car_tn' not found
+
+# create plumber.R for tabnet
+
+    Code
+      cat(readr::read_lines(tmp), sep = "\n")
+    Output
+      # Generated by the vetiver package; edit with care
+      
+      library(pins)
+      library(plumber)
+      library(rapidoc)
+      library(vetiver)
+      
+      # Packages needed to generate model predictions
+      if (FALSE) {
+          library(tabnet)
+      }
+      b <- board_folder(path = "<redacted>")
+      v <- vetiver_pin_read(b, "cars3")
+      
+      #* @plumber
+      function(pr) {
+          pr %>% vetiver_api(v)
+      }
+
diff --git a/tests/testthat/test-tabnet.R b/tests/testthat/test-tabnet.R
new file mode 100644
index 00000000..9209bade
--- /dev/null
+++ b/tests/testthat/test-tabnet.R
@@ -0,0 +1,74 @@
+library(pins)
+library(plumber)
+skip_if_not_installed("tabnet")
+
+set.seed(321)
+cars_tn <- tabnet::tabnet_fit(mpg ~ ., data = mtcars, epoch=30)
+v <- vetiver_model(cars_tn, "cars3", ptype_data = mtcars[,-1])
+
+test_that("can print tabnet model", {
+    expect_snapshot(v)
+})
+
+test_that("error for no ptype_data with tabnet", {
+    expect_snapshot(vetiver_model(car_tn, "cars3"), error = TRUE)
+})
+
+test_that("can predict tabnet model", {
+    preds <- predict(v, mtcars[,-1])
+    expect_equal(length(preds$.pred), 32)
+    expect_equal(mean(preds$.pred), 44, tolerance = 0.1)
+})
+
+
+test_that("can pin an tabnet model", {
+    b <- board_temp()
+    vetiver_pin_write(b, v)
+    pinned <- pin_read(b, "cars3")
+    expect_equal(pinned$model, butcher::butcher(cars_tn))
+    expect_equal(
+        pinned$ptype,
+        vctrs::vec_slice(tibble::as_tibble(mtcars[,-1]), 0)
+    )
+    expect_equal(
+        pinned$required_pkgs,
+        "tabnet"
+    )
+})
+
+test_that("default endpoint for tabnet", {
+    p <- pr() %>% vetiver_api(v)
+    p_routes <- p$routes[-1]
+    expect_equal(names(p_routes), c("ping", "predict"))
+    expect_equal(purrr::map_chr(p_routes, "verbs"),
+                 c(ping = "GET", predict = "POST"))
+})
+
+test_that("default OpenAPI spec", {
+    v$metadata <- list(url = "potatoes")
+    p <- pr() %>% vetiver_api(v)
+    car_spec <- p$getApiSpec()
+    expect_equal(car_spec$info$description,
+                 "A tabnet `nn_module` containing 6,301 parameters.")
+    post_spec <- car_spec$paths$`/predict`$post
+    expect_equal(names(post_spec), c("summary", "requestBody", "responses"))
+    expect_equal(as.character(post_spec$summary),
+                 "Return predictions from model using 10 features")
+    get_spec <- car_spec$paths$`/pin-url`$get
+    expect_equal(as.character(get_spec$summary),
+                 "Get URL of pinned vetiver model")
+
+})
+
+test_that("create plumber.R for tabnet", {
+    skip_on_cran()
+    b <- board_folder(path = tmp_dir)
+    vetiver_pin_write(b, v)
+    tmp <- tempfile()
+    vetiver_write_plumber(b, "cars3", file = tmp)
+    expect_snapshot(
+        cat(readr::read_lines(tmp), sep = "\n"),
+        transform = redact_vetiver
+    )
+})
+