fix(learner): printer uses cli

DESCRIPTION

@@ -46,6 +46,7 @@ Depends:
     R (>= 3.5.0)
 Imports:
     backports,
+    cli,
     checkmate (>= 2.2.0),
     data.table,
     lgr,

R/LearnerTorch.R

@@ -262,9 +262,11 @@ LearnerTorch = R6Class("LearnerTorch",
     #'   Currently unused.
     print = function(...) {
       super$print(...)
-      catn(str_indent("* Optimizer:", private$.optimizer$id))
-      catn(str_indent("* Loss:", private$.loss$id))
-      catn(str_indent("* Callbacks:", if (length(private$.callbacks)) as_short_string(paste0(ids(private$.callbacks), collapse = ","), 1000L) else "-"))
+      mlr3misc::cat_cli({
+        cli::cli_li("Optimizer: {private$.optimizer$id}")
+        cli::cli_li("Loss: {private$.loss$id}")
+        cli::cli_li(paste0("Callbacks: ", if (length(private$.callbacks)) as_short_string(paste0(ids(private$.callbacks), collapse = ","), 1000L) else "-"))
+      })
     },
     #' @description
     #' Marshal the learner.

tests/testthat/_snaps/LearnerTorch.md

@@ -0,0 +1,23 @@
+# printer
+
+    Code
+      lrn("classif.mlp", callbacks = list(t_clbk("history"), t_clbk("progress")))
+    Output
+
+      -- <LearnerTorchMLP> (classif.mlp): Multi Layer Perceptron ---------------------
+      * Model: -
+      * Parameters: device=auto, num_threads=1, num_interop_threads=1, seed=random,
+      eval_freq=1, measures_train=<list>, measures_valid=<list>, patience=0,
+      min_delta=0, shuffle=TRUE, tensor_dataset=FALSE, jit_trace=FALSE,
+      neurons=integer(0), p=0.5, activation=<nn_relu>, activation_args=<list>
+      * Validate: NULL
+      * Packages: mlr3, mlr3torch, torch, and progress
+      * Predict Types: [response] and prob
+      * Feature Types: integer, numeric, and lazy_tensor
+      * Encapsulation: none (fallback: -)
+      * Properties: internal_tuning, marshal, multiclass, twoclass, and validation
+      * Other settings: use_weights = 'error'
+      * Optimizer: adam
+      * Loss: cross_entropy
+      * Callbacks: history,progress
+

tests/testthat/test_LearnerTorch.R

@@ -1042,3 +1042,8 @@ test_that("NA prediction during validation does not cause issues.", {
     is.na(learner$model$callbacks$history$valid.regr.mse[1L]),
   )
 })
+
+test_that("printer", {
+  expect_snapshot(lrn("classif.mlp",
+    callbacks = list(t_clbk("history"), t_clbk("progress"))))
+})