different hash and man

Author: cxzhang4

R/TaskClassif_cifar.R

@@ -149,8 +149,8 @@ load_task_cifar10 = function(id = "cifar10") {
 
   task$col_roles$feature = "image"
 
-  # TODO: different hash, same manual
-  backend$hash = task$man = "mlr3torch::mlr_tasks_cifar"
+  backend$hash = "mlr3torch::mlr_tasks_cifar10"
+  task$man = "mlr3torch::mlr_tasks_cifar"
 
   task$filter(1:50000)
 
@@ -253,7 +253,8 @@ load_task_cifar100 = function(id = "cifar100") {
 
   task$col_roles$feature = "image"
 
-  backend$hash = task$man = "mlr3torch::mlr_tasks_cifar"
+  backend$hash = "mlr3torch::mlr_tasks_cifar100"
+  task$man = "mlr3torch::mlr_tasks_cifar"
 
   task$filter(1:50000)
 

tests/testthat/test_TaskClassif_cifar.R

@@ -19,59 +19,21 @@ test_that("CIFAR-10 works", {
   expect_equal(task$backend$ncol, 6)
 })
 
-# TODO: delete
-test_that("CIFAR-10 data matches the torchvision implementation", {
+test_that("CIFAR-100 works", {
   withr::local_options(mlr3torch.cache = TRUE)
-  task = tsk("cifar10")
-  task$data()
-
-  # check the responses
-  train_idx = 1:50000
-  int_mlr3torch_responses = as.integer(task$data()$class[train_idx])
-
-  cifar10_ds_train = cifar10_dataset(root = file.path(get_cache_dir(), "datasets", "cifar10", "raw"), train = TRUE,
-    download = FALSE)
-
-  # TODO: determine whether a separate function is truly necessary
-  # proably not, if getitem() allows a vector
-  get_response = function(idx, ds) {
-    ds$.getitem(idx)$y
-  }
-  cifar10_ds_responses = map_int(train_idx, get_response, ds = cifar10_ds_train)
-
-  expect_true(all.equal(int_mlr3torch_responses, cifar10_ds_responses))
-  
-  # check a subset of train images
-  small_train_idx = c(1, 2, 27, 9999,
-    10000, 10001, 10901, 19999,
-    20000, 20001, 29999,
-    30000, 30001, 39999,
-    40000, 40001, 49999,
-    50000
-  )
-  task_small = task$clone()
-  task_small$filter(small_train_idx)
-
-  test_same_at_idx = function(idx, lt_list, imgs_arr) {
-    all.equal(as.array(lt_list[[idx]]), imgs_arr[idx, , , ])
-  }
-
-  lt_list = materialize(task_small$data()$image)
-  imgs_arr = cifar10_ds_train$.getitem(small_train_idx)$x
-
-  expect_true(all(map_lgl(1:length(small_train_idx), test_same_at_idx, lt_list = lt_list, imgs_arr = imgs_arr)))
-
-  # check a subset of test images
-  test_idx = c(1, 2, 27, 8484, 9999, 10000)
-  
-  test_dt_from_task = task$backend$data(rows = 50001:60000, cols = task$backend$colnames)
-  expect_true(all(test_dt_from_task$split == "test"))
+  task = tsk("cifar100")
 
-  lt_list_test = materialize(test_dt_from_task[test_idx, ]$image)
-
-  cifar10_ds_test = cifar10_dataset(root = file.path(get_cache_dir(), "datasets", "cifar10", "raw"), train = FALSE,
-    download = FALSE)
-  imgs_arr_test = cifar10_ds_test$.getitem(test_idx)$x
+  expect_equal(task$nrow, 50000)
 
-  expect_true(all(map_lgl(1:length(test_idx), test_same_at_idx, lt_list = lt_list, imgs_arr = imgs_arr)))
+  task$filter(1:10)
+  expect_equal(task$id, "cifar100")
+  expect_equal(task$label, "CIFAR-100 Classification")
+  expect_equal(task$feature_names, "image")
+  expect_equal(task$target_names, "class")
+  expect_equal(task$man, "mlr3torch::mlr_tasks_cifar10")
+  task$data()
+  expect_true("cifar-100-binary" %in% list.files(file.path(get_cache_dir(), "datasets", "cifar100", "raw")))
+  expect_true("data.rds" %in% list.files(file.path(get_cache_dir(), "datasets", "cifar100")))
+  expect_equal(task$backend$nrow, 60000)
+  expect_equal(task$backend$ncol, 6)
 })