diff --git a/R/constraints.R b/R/constraints.R
index 8533ae6..400801e 100644
--- a/R/constraints.R
+++ b/R/constraints.R
@@ -23,74 +23,44 @@
 #' @seealso [luz_model_sequential()]
 #'
 #' @examples
-#' # Generate random data:
-#' train_x = matrix(rnorm(200), ncol=2)
-#' train_y = rnorm(100)
-#' # ---- Example with a keras/tensorflow network ----
-#' # Build a small nn:
-#' nn <- keras::keras_model_sequential()
-#' nn <- keras::layer_dense(nn, units = 10, activation = "tanh", input_shape = 2)
-#' nn <- keras::layer_dense(nn, units = 1, activation = "linear")
-#'
-#' # Add constraints
-#' nn_constrained <- add_constraints(nn, constraint_type = "l1_norm")
-#'
-#' # Compile and train (2 epochs )
-#' keras::compile(nn_constrained,
-#'                loss = "mse",
-#'                optimizer = keras::optimizer_adam(),
-#'                metrics = "mse")
-#'
-#' history <- keras::fit(nn_constrained,
-#'                       train_x,
-#'                       train_y,
-#'                       verbose = 0,
-#'                       epochs = 2,
-#'                       batch_size = 50,
-#'                       validation_split = 0.2
-#' )
-#'
-#' #' # ---- Example with a luz/torch network ----
-#' # Create a torch data loader to be able to train the NN:
-#' # Divide in only train and validation
-#' all_indices   <- 1:nrow(train_x)
-#' only_train_indices <- sample(all_indices, size = round(nrow(train_x)) * 0.8)
-#' val_indices   <- setdiff(all_indices, only_train_indices)
-#'
-#' # Create lists with x and y values to feed luz::as_dataloader()
-#' only_train_x <- as.matrix(train_x[only_train_indices,])
-#' only_train_y <- as.matrix(train_y[only_train_indices])
-#' val_x <- as.matrix(train_x[val_indices,])
-#' val_y <- as.matrix(train_y[val_indices])
-#'
-#' only_train_list <- list(x = only_train_x, y = only_train_y)
-#' val_list <- list(x = val_x, y = val_y)
-#'
-#' torch_data <- list(
-#'   train = luz::as_dataloader(only_train_list, batch_size = 50, shuffle = TRUE),
-#'   valid = luz::as_dataloader(val_list, batch_size = 50)
-#' )
-#'
-#' # Build a small nn:
-#' nn <- luz_model_sequential(
-#'         torch::nn_linear(2,10),
-#'         torch::nn_tanh(),
-#'         torch::nn_linear(10,1)
-#' )
-#'
-#' # Train the model adding the constraints inside the pipe
-#' # (equivalent to the approach used in the previous example)
-#' fitted <- nn |>
-#'    luz::setup(
-#'      loss = torch::nn_mse_loss(),
-#'      optimizer = torch::optim_adam,
-#'      metrics = list(
-#'        luz::luz_metric_mse()
-#'      )
-#'    ) |>
-#'    luz::fit(torch_data$train, epochs = 2, valid_data = torch_data$valid)
-#'
-#'
+#' \dontrun{
+#' if (requireNamespace("keras", quietly=TRUE)) {
+#'   # ---- Example with a keras/tensorflow network ----
+#'   # Build a small nn:
+#'   nn <- keras::keras_model_sequential()
+#'   nn <- keras::layer_dense(nn, units = 10, activation = "tanh", input_shape = 2)
+#'   nn <- keras::layer_dense(nn, units = 1, activation = "linear")
+#'
+#'   # Add constraints
+#'   nn_constrained <- add_constraints(nn, constraint_type = "l1_norm")
+#'
+#'   # Check that class of the constrained nn is "nn2poly"
+#'   class(nn_constrained)[1]
+#' }
+#' }
+#'
+#' if (requireNamespace("luz", quietly=TRUE)) {
+#'   # ---- Example with a luz/torch network ----
+#'
+#'   # Build a small nn
+#'   nn <- luz_model_sequential(
+#'     torch::nn_linear(2,10),
+#'     torch::nn_tanh(),
+#'     torch::nn_linear(10,1)
+#'   )
+#'
+#'   # With luz/torch we need to setup the nn before adding the constraints
+#'   nn <- luz::setup(module = nn,
+#'     loss = torch::nn_mse_loss(),
+#'     optimizer = torch::optim_adam,
+#'   )
+#'
+#'   # Add constraints
+#'   nn <- add_constraints(nn)
+#'
+#'   # Check that class of the constrained nn is "nn2poly"
+#'   class(nn)[1]
+#' }
 #'
 #' @export
 add_constraints <- function(object, type = c("l1_norm", "l2_norm"), ...) {
diff --git a/R/helpers.R b/R/helpers.R
index 18dcd6b..6d040f3 100644
--- a/R/helpers.R
+++ b/R/helpers.R
@@ -15,6 +15,7 @@
 #' @seealso [add_constraints()]
 #'
 #' @examples
+#' if (requireNamespace("luz", quietly=TRUE)) {
 #' # Create a NN using luz/torch as a sequential model
 #' # with 3 fully connected linear layers,
 #' # the first one with input = 5 variables,
@@ -31,6 +32,10 @@
 #'
 #' nn
 #'
+#' # Check that the nn is of class nn_squential
+#' class(nn)
+#' }
+#'
 #'
 #' @export
 luz_model_sequential <- function(...) {
diff --git a/man/add_constraints.Rd b/man/add_constraints.Rd
index ee0944b..e165348 100644
--- a/man/add_constraints.Rd
+++ b/man/add_constraints.Rd
@@ -32,74 +32,42 @@ Models in \code{luz/torch} need to use the \code{\link{luz_model_sequential}}
 helper in order to have a sequential model in the appropriate form.
 }
 \examples{
-# Generate random data:
-train_x = matrix(rnorm(200), ncol=2)
-train_y = rnorm(100)
-# ---- Example with a keras/tensorflow network ----
-# Build a small nn:
-nn <- keras::keras_model_sequential()
-nn <- keras::layer_dense(nn, units = 10, activation = "tanh", input_shape = 2)
-nn <- keras::layer_dense(nn, units = 1, activation = "linear")
+\dontrun{
+  # ---- Example with a keras/tensorflow network ----
+  # Build a small nn:
+  nn <- keras::keras_model_sequential()
+  nn <- keras::layer_dense(nn, units = 10, activation = "tanh", input_shape = 2)
+  nn <- keras::layer_dense(nn, units = 1, activation = "linear")
 
-# Add constraints
-nn_constrained <- add_constraints(nn, constraint_type = "l1_norm")
+  # Add constraints
+  nn_constrained <- add_constraints(nn, constraint_type = "l1_norm")
 
-# Compile and train (2 epochs )
-keras::compile(nn_constrained,
-               loss = "mse",
-               optimizer = keras::optimizer_adam(),
-               metrics = "mse")
-
-history <- keras::fit(nn_constrained,
-                      train_x,
-                      train_y,
-                      verbose = 0,
-                      epochs = 2,
-                      batch_size = 50,
-                      validation_split = 0.2
-)
-
-#' # ---- Example with a luz/torch network ----
-# Create a torch data loader to be able to train the NN:
-# Divide in only train and validation
-all_indices   <- 1:nrow(train_x)
-only_train_indices <- sample(all_indices, size = round(nrow(train_x)) * 0.8)
-val_indices   <- setdiff(all_indices, only_train_indices)
-
-# Create lists with x and y values to feed luz::as_dataloader()
-only_train_x <- as.matrix(train_x[only_train_indices,])
-only_train_y <- as.matrix(train_y[only_train_indices])
-val_x <- as.matrix(train_x[val_indices,])
-val_y <- as.matrix(train_y[val_indices])
-
-only_train_list <- list(x = only_train_x, y = only_train_y)
-val_list <- list(x = val_x, y = val_y)
+  # Check that class of the constrained nn is "nn2poly"
+  class(nn_constrained)[1]
+}
 
-torch_data <- list(
-  train = luz::as_dataloader(only_train_list, batch_size = 50, shuffle = TRUE),
-  valid = luz::as_dataloader(val_list, batch_size = 50)
-)
+if (requireNamespace("luz", quietly=TRUE)) {
+  # ---- Example with a luz/torch network ----
 
-# Build a small nn:
-nn <- luz_model_sequential(
-        torch::nn_linear(2,10),
-        torch::nn_tanh(),
-        torch::nn_linear(10,1)
-)
+  # Build a small nn
+  nn <- luz_model_sequential(
+    torch::nn_linear(2,10),
+    torch::nn_tanh(),
+    torch::nn_linear(10,1)
+  )
 
-# Train the model adding the constraints inside the pipe
-# (equivalent to the approach used in the previous example)
-fitted <- nn |>
-   luz::setup(
-     loss = torch::nn_mse_loss(),
-     optimizer = torch::optim_adam,
-     metrics = list(
-       luz::luz_metric_mse()
-     )
-   ) |>
-   luz::fit(torch_data$train, epochs = 2, valid_data = torch_data$valid)
+  # With luz/torch we need to setup the nn before adding the constraints
+  nn <- luz::setup(module = nn,
+    loss = torch::nn_mse_loss(),
+    optimizer = torch::optim_adam,
+  )
 
+  # Add constraints
+  nn <- add_constraints(nn)
 
+  # Check that class of the constrained nn is "nn2poly"
+  class(nn)[1]
+}
 
 }
 \seealso{
diff --git a/man/luz_model_sequential.Rd b/man/luz_model_sequential.Rd
index 31ffaa6..3a4f244 100644
--- a/man/luz_model_sequential.Rd
+++ b/man/luz_model_sequential.Rd
@@ -23,6 +23,7 @@ Furthermore, this step is also needed to be able to impose the needed
 constraints when using the \code{luz/torch} framework.
 }
 \examples{
+if (requireNamespace("luz", quietly=TRUE)) {
 # Create a NN using luz/torch as a sequential model
 # with 3 fully connected linear layers,
 # the first one with input = 5 variables,
@@ -39,6 +40,10 @@ nn <- luz_model_sequential(
 
 nn
 
+# Check that the nn is of class nn_squential
+class(nn)
+}
+
 
 }
 \seealso{