Merge pull request #2145 from recommenders-team/miguel/mind

Add new URL of MIND small and MIND large
recommenders-team · Aug 22, 2024 · fd5f861 · fd5f861
2 parents 61568e6 + 608f684
commit fd5f861
Show file tree

Hide file tree

Showing 3 changed files with 21 additions and 21 deletions.
diff --git a/recommenders/datasets/mind.py b/recommenders/datasets/mind.py
@@ -18,16 +18,16 @@
 
 
 URL_MIND_LARGE_TRAIN = (
-    "https://mind201910small.blob.core.windows.net/release/MINDlarge_train.zip"
+    "https://recodatasets.z20.web.core.windows.net/newsrec/MINDlarge_train.zip"
 )
 URL_MIND_LARGE_VALID = (
-    "https://mind201910small.blob.core.windows.net/release/MINDlarge_dev.zip"
+    "https://recodatasets.z20.web.core.windows.net/newsrec/MINDlarge_dev.zip"
 )
 URL_MIND_SMALL_TRAIN = (
-    "https://mind201910small.blob.core.windows.net/release/MINDsmall_train.zip"
+    "https://recodatasets.z20.web.core.windows.net/newsrec/MINDsmall_train.zip"
 )
 URL_MIND_SMALL_VALID = (
-    "https://mind201910small.blob.core.windows.net/release/MINDsmall_dev.zip"
+    "https://recodatasets.z20.web.core.windows.net/newsrec/MINDsmall_dev.zip"
 )
 URL_MIND_DEMO_TRAIN = (
     "https://recodatasets.z20.web.core.windows.net/newsrec/MINDdemo_train.zip"

diff --git a/recommenders/models/newsrec/newsrec_utils.py b/recommenders/models/newsrec/newsrec_utils.py
@@ -310,15 +310,15 @@ def get_mind_data_set(type):
 
     if type == "large":
         return (
-            "https://mind201910small.blob.core.windows.net/release/",
+            "https://recodatasets.z20.web.core.windows.net/newsrec/",
             "MINDlarge_train.zip",
             "MINDlarge_dev.zip",
             "MINDlarge_utils.zip",
         )
 
     elif type == "small":
         return (
-            "https://mind201910small.blob.core.windows.net/release/",
+            "https://recodatasets.z20.web.core.windows.net/newsrec/",
             "MINDsmall_train.zip",
             "MINDsmall_dev.zip",
             "MINDsmall_utils.zip",

diff --git a/tests/data_validation/recommenders/datasets/test_mind.py b/tests/data_validation/recommenders/datasets/test_mind.py
@@ -27,34 +27,34 @@
             '"0x8D8B8AD5B126C3B"',
         ),
         (
-            "https://mind201910small.blob.core.windows.net/release/MINDsmall_train.zip",
-            "52952752",
+            "https://recodatasets.z20.web.core.windows.net/newsrec/MINDsmall_train.zip",
+            "52953372",
             "0x8D834F2EB31BDEC",
         ),
         (
-            "https://mind201910small.blob.core.windows.net/release/MINDsmall_dev.zip",
-            "30945572",
+            "https://recodatasets.z20.web.core.windows.net/newsrec/MINDsmall_dev.zip",
+            "30946172",
             "0x8D834F2EBA8D865",
         ),
         (
-            "https://mind201910small.blob.core.windows.net/release/MINDsmall_utils.zip",
+            "https://recodatasets.z20.web.core.windows.net/newsrec/MINDsmall_utils.zip",
             "155178106",
             "0x8D87F67F4AEB960",
         ),
         (
-            "https://mind201910small.blob.core.windows.net/release/MINDlarge_train.zip",
-            "530196631",
+            "https://recodatasets.z20.web.core.windows.net/newsrec/MINDlarge_train.zip",
+            "531361237",
             "0x8D8244E90C15C07",
         ),
         (
-            "https://mind201910small.blob.core.windows.net/release/MINDlarge_dev.zip",
-            "103456245",
+            "https://recodatasets.z20.web.core.windows.net/newsrec/MINDlarge_dev.zip",
+            "103593383",
             "0x8D8244E92005849",
         ),
         (
-            "https://mind201910small.blob.core.windows.net/release/MINDlarge_utils.zip",
+            "https://recodatasets.z20.web.core.windows.net/newsrec/MINDlarge_utils.zip",
             "150359301",
-            "0x8D87F67E6CA4364",
+            "0x8D8B8AD5B2ED4C9",
         ),
     ],
 )
@@ -75,9 +75,9 @@ def test_download_mind_demo(tmp):
 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 == 52952752
+    assert statinfo.st_size == 52953372
     statinfo = os.stat(valid_path)
-    assert statinfo.st_size == 30945572
+    assert statinfo.st_size == 30946172
 
 
 def test_extract_mind_demo(tmp):
@@ -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 == 530196631
+    assert statinfo.st_size == 531361237
     statinfo = os.stat(valid_path)
-    assert statinfo.st_size == 103456245
+    assert statinfo.st_size == 103593383
 
 
 def test_extract_mind_large(tmp):