pytorch · parmeet · May 27, 2022 · May 27, 2022
diff --git a/test/datasets/test_agnews.py b/test/datasets/test_agnews.py
@@ -42,7 +42,7 @@ class TestAGNews(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_amazonreviews.py b/test/datasets/test_amazonreviews.py
@@ -69,7 +69,9 @@ def tearDownClass(cls):
 
     @nested_params([AmazonReviewFull, AmazonReviewPolarity], ["train", "test"])
     def test_amazon_reviews(self, amazon_review_dataset, split):
-        expected_samples = _get_mock_dataset(self.root_dir, amazon_review_dataset.__name__)[split]
+        expected_samples = _get_mock_dataset(os.path.join(self.root_dir, "datasets"), amazon_review_dataset.__name__)[
+            split
+        ]
         dataset = amazon_review_dataset(root=self.root_dir, split=split)
         samples = list(dataset)
 

diff --git a/test/datasets/test_cc100.py b/test/datasets/test_cc100.py
@@ -40,7 +40,7 @@ class TestCC100(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_cola.py b/test/datasets/test_cola.py
@@ -51,7 +51,7 @@ class TestCoLA(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_conll2000chunking.py b/test/datasets/test_conll2000chunking.py
@@ -53,7 +53,7 @@ class TestCoNLL2000Chunking(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_dbpedia.py b/test/datasets/test_dbpedia.py
@@ -50,7 +50,7 @@ class TestDBpedia(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_enwik9.py b/test/datasets/test_enwik9.py
@@ -47,7 +47,7 @@ class TestEnWik9(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_imdb.py b/test/datasets/test_imdb.py
@@ -56,7 +56,7 @@ class TestIMDB(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_iwslt2016.py b/test/datasets/test_iwslt2016.py
@@ -164,7 +164,7 @@ def tearDownClass(cls):
     def test_iwslt2016(self, split, src, tgt, dev_set, test_set):
 
         with tempfile.TemporaryDirectory() as root_dir:
-            expected_samples = _get_mock_dataset(root_dir, split, src, tgt, dev_set, test_set)
+            expected_samples = _get_mock_dataset(os.path.join(root_dir, "datasets"), split, src, tgt, dev_set, test_set)
 
             dataset = IWSLT2016(
                 root=root_dir,
@@ -185,7 +185,9 @@ def test_iwslt2016_split_argument(self, split):
             language_pair = ("de", "en")
             valid_set = "tst2013"
             test_set = "tst2014"
-            _ = _get_mock_dataset(root_dir, split, language_pair[0], language_pair[1], valid_set, test_set)
+            _ = _get_mock_dataset(
+                os.path.join(root_dir, "datasets"), split, language_pair[0], language_pair[1], valid_set, test_set
+            )
             dataset1 = IWSLT2016(
                 root=root_dir,
                 split=split,

diff --git a/test/datasets/test_iwslt2017.py b/test/datasets/test_iwslt2017.py
@@ -155,7 +155,9 @@ def tearDownClass(cls):
     def test_iwslt2017(self, split, src, tgt):
 
         with tempfile.TemporaryDirectory() as root_dir:
-            expected_samples = _get_mock_dataset(root_dir, split, src, tgt, "dev2010", "tst2010")
+            expected_samples = _get_mock_dataset(
+                os.path.join(root_dir, "datasets"), split, src, tgt, "dev2010", "tst2010"
+            )
 
             dataset = IWSLT2017(root=root_dir, split=split, language_pair=(src, tgt))
 
@@ -170,7 +172,9 @@ def test_iwslt2017_split_argument(self, split):
             language_pair = ("de", "en")
             valid_set = "dev2010"
             test_set = "tst2010"
-            _ = _get_mock_dataset(root_dir, split, language_pair[0], language_pair[1], valid_set, test_set)
+            _ = _get_mock_dataset(
+                os.path.join(root_dir, "datasets"), split, language_pair[0], language_pair[1], valid_set, test_set
+            )
             dataset1 = IWSLT2017(root=root_dir, split=split, language_pair=language_pair)
             (dataset2,) = IWSLT2017(root=root_dir, split=(split,), language_pair=language_pair)
 

diff --git a/test/datasets/test_mnli.py b/test/datasets/test_mnli.py
@@ -56,7 +56,7 @@ class TestMNLI(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_mrpc.py b/test/datasets/test_mrpc.py
@@ -44,7 +44,7 @@ class TestMRPC(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_multi30k.py b/test/datasets/test_multi30k.py
@@ -47,7 +47,7 @@ class TestMulti30k(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_penntreebank.py b/test/datasets/test_penntreebank.py
@@ -41,7 +41,7 @@ class TestPennTreebank(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_qqp.py b/test/datasets/test_qqp.py
@@ -41,7 +41,7 @@ class TestQQP(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_sogounews.py b/test/datasets/test_sogounews.py
@@ -48,7 +48,7 @@ class TestSogouNews(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_squads.py b/test/datasets/test_squads.py
@@ -87,7 +87,7 @@ def tearDownClass(cls):
 
     @nested_params([SQuAD1, SQuAD2], ["train", "dev"])
     def test_squads(self, squad_dataset, split):
-        expected_samples = _get_mock_dataset(self.root_dir, squad_dataset.__name__)[split]
+        expected_samples = _get_mock_dataset(os.path.join(self.root_dir, "datasets"), squad_dataset.__name__)[split]
         dataset = squad_dataset(root=self.root_dir, split=split)
         samples = list(dataset)
 

diff --git a/test/datasets/test_sst2.py b/test/datasets/test_sst2.py
@@ -59,7 +59,7 @@ class TestSST2(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_stsb.py b/test/datasets/test_stsb.py
@@ -62,7 +62,7 @@ class TestSTSB(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_udpos.py b/test/datasets/test_udpos.py
@@ -56,7 +56,7 @@ class TestUDPOS(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_wikitexts.py b/test/datasets/test_wikitexts.py
@@ -71,7 +71,9 @@ def tearDownClass(cls):
 
     @nested_params([WikiText103, WikiText2], ["train", "valid", "test"])
     def test_wikitexts(self, wikitext_dataset, split):
-        expected_samples = _get_mock_dataset(self.root_dir, base_dir_name=wikitext_dataset.__name__)[split]
+        expected_samples = _get_mock_dataset(
+            os.path.join(self.root_dir, "datasets"), base_dir_name=wikitext_dataset.__name__
+        )[split]
 
         dataset = wikitext_dataset(root=self.root_dir, split=split)
         samples = list(dataset)

diff --git a/test/datasets/test_yahooanswers.py b/test/datasets/test_yahooanswers.py
@@ -48,7 +48,7 @@ class TestYahooAnswers(TempDirMixin, TorchtextTestCase):
     def setUpClass(cls):
         super().setUpClass()
         cls.root_dir = cls.get_base_temp_dir()
-        cls.samples = _get_mock_dataset(cls.root_dir)
+        cls.samples = _get_mock_dataset(os.path.join(cls.root_dir, "datasets"))
         cls.patcher = patch("torchdata.datapipes.iter.util.cacheholder._hash_check", return_value=True)
         cls.patcher.start()
 

diff --git a/test/datasets/test_yelpreviews.py b/test/datasets/test_yelpreviews.py
@@ -70,7 +70,9 @@ def tearDownClass(cls):
 
     @nested_params([YelpReviewPolarity, YelpReviewFull], ["train", "test"])
     def test_yelpreviews(self, yelp_dataset, split):
-        expected_samples = _get_mock_dataset(self.root_dir, base_dir_name=yelp_dataset.__name__)[split]
+        expected_samples = _get_mock_dataset(
+            os.path.join(self.root_dir, "datasets"), base_dir_name=yelp_dataset.__name__
+        )[split]
 
         dataset = yelp_dataset(root=self.root_dir, split=split)
         samples = list(dataset)