Added PT 2.0 SM updates

Author: Mike Schneider

src/sagemaker/fw_utils.py

@@ -135,6 +135,7 @@
         "1.12.0",
         "1.12.1",
         "1.13.1",
+        "2.0.0",
     ],
 }
 
@@ -148,6 +149,7 @@
     "1.12.0",
     "1.12.1",
     "1.13.1",
+    "2.0.0",
 ]
 
 
@@ -161,6 +163,7 @@
     "1.12.0",
     "1.12.1",
     "1.13.1",
+    "2.0.0",
 ]
 
 SMDISTRIBUTED_SUPPORTED_STRATEGIES = ["dataparallel", "modelparallel"]

src/sagemaker/image_uri_config/pytorch.json

@@ -992,7 +992,8 @@
             "1.10": "1.10.2",
             "1.11": "1.11.0",
             "1.12": "1.12.1",
-            "1.13": "1.13.1"
+            "1.13": "1.13.1",
+            "2.0": "2.0.0"
         },
         "versions": {
             "0.4.0": {
@@ -1754,6 +1755,43 @@
                     "us-west-2": "763104351884"
                 },
                 "repository": "pytorch-training"
+            },
+            "2.0.0": {
+                "py_versions": [
+                    "py310"
+                ],
+                "registries": {
+                    "af-south-1": "626614931356",
+                    "ap-east-1": "871362719292",
+                    "ap-northeast-1": "763104351884",
+                    "ap-northeast-2": "763104351884",
+                    "ap-northeast-3": "364406365360",
+                    "ap-south-1": "763104351884",
+                    "ap-southeast-1": "763104351884",
+                    "ap-southeast-2": "763104351884",
+                    "ap-southeast-3": "907027046896",
+                    "ap-southeast-4": "457447274322",
+                    "ca-central-1": "763104351884",
+                    "cn-north-1": "727897471807",
+                    "cn-northwest-1": "727897471807",
+                    "eu-central-1": "763104351884",
+                    "eu-north-1": "763104351884",
+                    "eu-west-1": "763104351884",
+                    "eu-west-2": "763104351884",
+                    "eu-west-3": "763104351884",
+                    "eu-south-1": "692866216735",
+                    "me-south-1": "217643126080",
+                    "sa-east-1": "763104351884",
+                    "us-east-1": "763104351884",
+                    "us-east-2": "763104351884",
+                    "us-gov-east-1": "446045086412",
+                    "us-gov-west-1": "442386744353",
+                    "us-iso-east-1": "886529160074",
+                    "us-isob-east-1": "094389454867",
+                    "us-west-1": "763104351884",
+                    "us-west-2": "763104351884"
+                },
+                "repository": "pytorch-training"
             }
         }
     }

tests/unit/test_fw_utils.py

@@ -913,6 +913,7 @@ def test_validate_smdataparallel_args_not_raises():
         ("ml.p3.16xlarge", "pytorch", "1.12.1", "py38", smdataparallel_enabled),
         ("ml.p3.16xlarge", "pytorch", "1.12", "py38", smdataparallel_enabled),
         ("ml.p3.16xlarge", "pytorch", "1.13.1", "py39", smdataparallel_enabled),
+        ("ml.p3.16xlarge", "pytorch", "2.0.0", "py310", smdataparallel_enabled),
         ("ml.p3.16xlarge", "tensorflow", "2.4.1", "py3", smdataparallel_enabled_custom_mpi),
         ("ml.p3.16xlarge", "tensorflow", "2.4.1", "py37", smdataparallel_enabled_custom_mpi),
         ("ml.p3.16xlarge", "tensorflow", "2.4.3", "py3", smdataparallel_enabled_custom_mpi),
@@ -934,6 +935,7 @@ def test_validate_smdataparallel_args_not_raises():
         ("ml.p3.16xlarge", "pytorch", "1.12.0", "py38", smdataparallel_enabled_custom_mpi),
         ("ml.p3.16xlarge", "pytorch", "1.12.1", "py38", smdataparallel_enabled_custom_mpi),
         ("ml.p3.16xlarge", "pytorch", "1.13.1", "py39", smdataparallel_enabled_custom_mpi),
+        ("ml.p3.16xlarge", "pytorch", "2.0.0", "py310", smdataparallel_enabled_custom_mpi),
     ]
     for instance_type, framework_name, framework_version, py_version, distribution in good_args:
         fw_utils._validate_smdataparallel_args(
@@ -1034,6 +1036,7 @@ def test_validate_torch_distributed_not_raises():
     torch_distributed_enabled = {"torch_distributed": {"enabled": True}}
     torch_distributed_gpu_supported_fw_versions = [
         "1.13.1",
+        "2.0.0",
     ]
     for framework_version in torch_distributed_gpu_supported_fw_versions:
         fw_utils.validate_torch_distributed_distribution(

tests/unit/test_utils.py

@@ -380,8 +380,8 @@ def test_set_nested_value():
 
 
 def test_get_short_version():
-    assert sagemaker.utils.get_short_version("1.13.1") == "1.13"
-    assert sagemaker.utils.get_short_version("1.13") == "1.13"
+    assert sagemaker.utils.get_short_version("2.0.0") == "2.0"
+    assert sagemaker.utils.get_short_version("2.0") == "2.0"
 
 
 def test_deferred_error():