small corrections

R/LearnerTorch.R

@@ -22,7 +22,8 @@
 #'   See [`mlr_reflections$learner_predict_types`][mlr_reflections] for available values.
 #'   For regression, the default is `"response"`.
 #'   For classification, this defaults to `"response"` and `"prob"`.
-#'   To deviate from the defaults, it is necessary to overwrite the private `$.predict()` method.
+#'   To deviate from the defaults, it is necessary to overwrite the private `$.encode_prediction()`
+#'   method, see section *Inheriting*.
 #' @param loss (`NULL` or [`TorchLoss`])\cr
 #'   The loss to use for training.
 #'   Defaults to MSE for regression and cross entropy for classification.
@@ -71,7 +72,7 @@
 #'   ([`torch_tensor`], [`Task`], `list()`) -> `list()`\cr
 #'   Take in the raw predictions from `self$network` (`predict_tensor`) and encode them into a
 #'   format that can be converted to valid `mlr3` predictions using [`mlr3::as_prediction_data()`].
-#'   This must take `self$predict_type` into account.
+#'   This method must take `self$predict_type` into account.
 #'
 #' While it is possible to add parameters by specifying the `param_set` construction argument, it is currently
 #' not possible to remove existing parameters, i.e. those listed in section *Parameters*.
@@ -88,15 +89,6 @@ LearnerTorch = R6Class("LearnerTorch",
     initialize = function(id, task_type, param_set, properties, man, label, feature_types,
       optimizer = NULL, loss = NULL, packages = NULL, predict_types = NULL, callbacks = list()) {
       assert_choice(task_type, c("regr", "classif"))
-      predict_types = predict_types %??% switch(task_type,
-        regr = "response",
-        classif = c("response", "prob")
-      )
-      loss = loss %??% switch(task_type,
-        classif = t_loss("cross_entropy"),
-        regr = t_loss("mse")
-      )
-      optimizer = optimizer %??% t_opt("adam")
 
       learner_torch_initialize(self = self, private = private, super = super,
         task_type = task_type,
@@ -118,6 +110,7 @@ LearnerTorch = R6Class("LearnerTorch",
     #' @field network ([`nn_module()`][torch::nn_module])\cr
     #'   The network (only available after training).
     network = function(rhs) {
+      assert_ro_binding(rhs)
       if (is.null(self$state)) {
         stopf("Cannot access network before training.")
       }

R/LearnerTorchAlexNet.R

@@ -27,7 +27,6 @@ LearnerTorchAlexNet = R6Class("LearnerTorchAlexNet",
         pretrained = p_lgl(tags = c("required", "train"))
       )
       param_set$values = list(pretrained = TRUE)
-      # TODO: Freezing --> maybe as a callback?
       super$initialize(
         task_type = task_type,
         id = paste0(task_type, ".alexnet"),

R/LearnerTorchFeatureless.R

@@ -31,7 +31,6 @@ LearnerTorchFeatureless = R6Class("LearnerTorchFeatureless",
         label = "Featureless Torch Learner",
         param_set = ps(),
         properties = properties,
-        # TODO: This should have all feature types, and have prop"),
         feature_types = unname(mlr_reflections$task_feature_types),
         man = "mlr3torch::mlr_learners.torch_featureless",
         optimizer = optimizer,

R/LearnerTorchImage.R

@@ -48,7 +48,7 @@ LearnerTorchImage = R6Class("LearnerTorchImage",
       callbacks = list(), packages = c("torchvision", "magick"), man, properties = NULL,
       predict_types = NULL) {
       properties = properties %??% switch(task_type,
-        regr = c(),
+        regr = character(0),
         classif = c("twoclass", "multiclass")
       )
       assert_param_set(param_set)

R/learner_torch_methods.R

@@ -26,6 +26,16 @@ learner_torch_initialize = function(
   label,
   callbacks
   ) {
+  predict_types = predict_types %??% switch(task_type,
+    regr = "response",
+    classif = c("response", "prob")
+  )
+  loss = loss %??% switch(task_type,
+    classif = t_loss("cross_entropy"),
+    regr = t_loss("mse")
+  )
+  optimizer = optimizer %??% t_opt("adam")
+
   private$.optimizer = as_torch_optimizer(optimizer, clone = TRUE)
   private$.optimizer$param_set$set_id = "opt"