Adding shuffled version of the toxicity dataset

CHANGES.rst

@@ -22,6 +22,8 @@ Changes
 - The row indices of training and testing samples are now also included in the
   dictionaries produced by :meth:`DataOp.skb.iter_cv_splits`. :pr:`2012` by
   :user:`Jérôme Dockès <jeromedockes>`.
+- :func:`fetch_toxicity_dataset` now returns a shuffled version of the dataset by default.
+  :pr:`1892` by user:`Riccardo Cappuzzo <rcap107>`.
 
 Bugfixes
 --------
@@ -133,11 +135,6 @@ Changes
   :pr:`1819` by :user:`Eloi Massoulié <emassoulie>`
 - :func:`compute_ngram_distance` has been renamed to :func:`_compute_ngram_distance` and is now a private function.
   :pr:`1838` by :user:`Siddharth Baleja <siddharthbaleja>`.
-- The repository wheel has been made smaller by removing some material that was
-  not necessary for using the library. Benchmarks are now available in a separate
-  `repository <https://github.com/skrub-data/skrub-benchmarks>`__.
-  :pr:`1893` by :user:`Riccardo Cappuzzo <rcap107>`.
-
 
 Bugfixes
 --------

skrub/datasets/_fetching.py

@@ -352,7 +352,7 @@ def fetch_toxicity(data_home=None):
         path : str
             The path to the toxicity CSV file.
     """
-    return load_simple_dataset("toxicity", data_home)
+    return load_simple_dataset("toxicity_shuffled", data_home)
 
 
 def fetch_videogame_sales(data_home=None):

skrub/datasets/_utils.py

@@ -106,6 +106,13 @@
         ],
         "sha256": "ee187c119925ea4cdb9abd7f0f3758159f042e71b172cafe5b784d79c7590ce3",
     },
+    "toxicity_shuffled": {
+        "urls": [
+            "https://github.com/skrub-data/skrub-data-files/raw/refs/heads/main/toxicity_shuffled.zip",
+            "https://osf.io/download/zebm7",
+        ],
+        "sha256": "01382d19987c04faab8c7b10dfa87719ae4af273dba08a48a95b0c9a69aeb009",
+    },
     "traffic_violations": {
         "urls": [
             "https://github.com/skrub-data/skrub-data-files/raw/refs/heads/main/traffic_violations.zip",

skrub/datasets/tests/test_fetching.py

@@ -85,7 +85,7 @@ def test_fetch_employee_salaries():
         ("midwest_survey", (2494, 29)),
         ("open_payments", (73558, 6)),
         ("traffic_violations", (1578154, 43)),
-        ("toxicity", (1000, 2)),
+        ("toxicity_shuffled", (1000, 2)),
         ("videogame_sales", (16572, 11)),
         ("bike_sharing", (17379, 11)),
     ],