Merge pull request #2155 from recommenders-team/miguel/mind2

Fix issue with MIND large and small
recommenders-team · Aug 26, 2024 · 4f5861d · 4f5861d
2 parents fd5f861 + fae2030
commit 4f5861d
Showing 1 changed file with 20 additions and 20 deletions.
diff --git a/tests/data_validation/recommenders/datasets/test_mind.py b/tests/data_validation/recommenders/datasets/test_mind.py
@@ -28,33 +28,33 @@
         ),
         (
             "https://recodatasets.z20.web.core.windows.net/newsrec/MINDsmall_train.zip",
-            "52953372",
-            "0x8D834F2EB31BDEC",
+            "52994575",
+            '"0x8DCC5A830190676"',
         ),
         (
             "https://recodatasets.z20.web.core.windows.net/newsrec/MINDsmall_dev.zip",
-            "30946172",
-            "0x8D834F2EBA8D865",
+            "30948560",
+            '"0x8DCC5A82E182A0F"',
         ),
         (
             "https://recodatasets.z20.web.core.windows.net/newsrec/MINDsmall_utils.zip",
             "155178106",
-            "0x8D87F67F4AEB960",
+            '"0x8D8B8AD5B3677C6"',
         ),
         (
             "https://recodatasets.z20.web.core.windows.net/newsrec/MINDlarge_train.zip",
-            "531361237",
-            "0x8D8244E90C15C07",
+            "531360717",
+            '"0x8DCC5A8375BDC1D"',
         ),
         (
             "https://recodatasets.z20.web.core.windows.net/newsrec/MINDlarge_dev.zip",
-            "103593383",
-            "0x8D8244E92005849",
+            "103592887",
+            '"0x8DCC5A82FE8609C"',
         ),
         (
             "https://recodatasets.z20.web.core.windows.net/newsrec/MINDlarge_utils.zip",
             "150359301",
-            "0x8D8B8AD5B2ED4C9",
+            '"0x8D8B8AD5B2ED4C9"',
         ),
     ],
 )
@@ -72,14 +72,6 @@ def test_download_mind_demo(tmp):
     assert statinfo.st_size == 10080022
 
 
-def test_download_mind_small(tmp):
-    train_path, valid_path = download_mind(size="small", dest_path=tmp)
-    statinfo = os.stat(train_path)
-    assert statinfo.st_size == 52953372
-    statinfo = os.stat(valid_path)
-    assert statinfo.st_size == 30946172
-
-
 def test_extract_mind_demo(tmp):
     train_zip, valid_zip = download_mind(size="demo", dest_path=tmp)
     train_path, valid_path = extract_mind(train_zip, valid_zip, clean_zip_file=False)
@@ -102,6 +94,14 @@ def test_extract_mind_demo(tmp):
     assert statinfo.st_size == 1044588
 
 
+def test_download_mind_small(tmp):
+    train_path, valid_path = download_mind(size="small", dest_path=tmp)
+    statinfo = os.stat(train_path)
+    assert statinfo.st_size == 52994575
+    statinfo = os.stat(valid_path)
+    assert statinfo.st_size == 30948560
+
+
 def test_extract_mind_small(tmp):
     train_zip, valid_zip = download_mind(size="small", dest_path=tmp)
     train_path, valid_path = extract_mind(train_zip, valid_zip, clean_zip_file=False)
@@ -127,9 +127,9 @@ def test_extract_mind_small(tmp):
 def test_download_mind_large(tmp_path):
     train_path, valid_path = download_mind(size="large", dest_path=tmp_path)
     statinfo = os.stat(train_path)
-    assert statinfo.st_size == 531361237
+    assert statinfo.st_size == 531360717
     statinfo = os.stat(valid_path)
-    assert statinfo.st_size == 103593383
+    assert statinfo.st_size == 103592887
 
 
 def test_extract_mind_large(tmp):