feat(learner): initialize shuffle to TRUE

NEWS.md

@@ -25,6 +25,7 @@
 * feat: `PipeOpBlock` should no longer create ID clashes with other PipeOps in the graph (#260)
 * fix: `device` is no longer part of the `dataset` which allows for parallel dataloading
   on GPUs.
+* feat: The `shuffle` parameter during model training is now initialized to `TRUE`.
 
 # mlr3torch 0.1.2
 

R/paramset_torchlearner.R

@@ -69,7 +69,7 @@ paramset_torchlearner = function(task_type) {
     min_delta             = p_dbl(lower = 0, tags = c("train", "required"), init = 0),
     # dataloader parameters
     batch_size            = p_int(tags = c("train", "predict", "required"), lower = 1L),
-    shuffle               = p_lgl(tags = "train", default = FALSE),
+    shuffle               = p_lgl(tags = "train", default = FALSE, init = TRUE),
     sampler               = p_uty(tags = c("train", "predict")),
     batch_sampler         = p_uty(tags = c("train", "predict")),
     num_workers           = p_int(lower = 0, default = 0, tags = c("train", "predict")),

man-roxygen/paramset_torchlearner.R

@@ -63,8 +63,8 @@
 #' * `batch_size` :: `integer(1)`\cr
 #'   The batch size (required).
 #' * `shuffle` :: `logical(1)`\cr
-#'   Whether to shuffle the instances in the dataset. Default is `FALSE`.
-#'   This does not impact validation.
+#'   Whether to shuffle the instances in the dataset. This is initialized to `TRUE`,
+#'   which differs from the default (`FALSE`).
 #' * `sampler` :: [`torch::sampler`]\cr
 #'   Object that defines how the dataloader draw samples.
 #' * `batch_sampler` :: [`torch::sampler`]\cr