Merge branch 'aws:master' into master

Author: cj-zhang

CHANGELOG.md

@@ -1,5 +1,59 @@
 # Changelog
 
+## v2.242.0 (2025-03-14)
+
+### Features
+
+ * add integ tests for training JumpStart models in private hub
+
+### Bug Fixes and Other Changes
+
+ * Torch upgrade
+ * Prevent RunContext overlap between test_run tests
+ * remove s3 output location requirement from hub class init
+ * Fixing Pytorch training python version in tests
+ * update image_uri_configs  03-11-2025 07:18:09 PST
+ * resolve infinite loop in _find_config on Windows systems
+ * pipeline definition function doc update
+
+## v2.241.0 (2025-03-06)
+
+### Features
+
+ * Make DistributedConfig Extensible
+ * support training for JumpStart model references as part of Curated Hub Phase 2
+ * Allow ModelTrainer to accept hyperparameters file
+
+### Bug Fixes and Other Changes
+
+ * Skip tests with deprecated instance type
+ * Ensure Model.is_repack() returns a boolean
+ * Fix error when there is no session to call _create_model_request()
+ * Use sagemaker session's s3_resource in download_folder
+ * Added check for the presence of model package group before creating one
+ * Fix key error in _send_metrics()
+
+## v2.240.0 (2025-02-25)
+
+### Features
+
+ * Add support for TGI Neuronx 0.0.27 and HF PT 2.3.0 image in PySDK
+
+### Bug Fixes and Other Changes
+
+ * Remove main function entrypoint in ModelBuilder dependency manager.
+ * forbid extras in Configs
+ * altconfig hubcontent and reenable integ test
+ * Merge branch 'master-rba' into local_merge
+ * py_version doc fixes
+ * Add backward compatbility for RecordSerializer and RecordDeserializer
+ * update image_uri_configs  02-21-2025 06:18:10 PST
+ * update image_uri_configs  02-20-2025 06:18:08 PST
+
+### Documentation Changes
+
+ * Removed a line about python version requirements of training script which can misguide users.
+
 ## v2.239.3 (2025-02-19)
 
 ### Bug Fixes and Other Changes

VERSION

@@ -1 +1 @@
-2.239.4.dev0
+2.242.1.dev0

doc/frameworks/pytorch/using_pytorch.rst

@@ -28,8 +28,6 @@ To train a PyTorch model by using the SageMaker Python SDK:
 Prepare a PyTorch Training Script
 =================================
 
-Your PyTorch training script must be a Python 3.6 compatible source file.
-
 Prepare your script in a separate source file than the notebook, terminal session, or source file you're
 using to submit the script to SageMaker via a ``PyTorch`` Estimator. This will be discussed in further detail below.
 

doc/overview.rst

@@ -30,6 +30,11 @@ To train a model by using the SageMaker Python SDK, you:
 
 After you train a model, you can save it, and then serve the model as an endpoint to get real-time inferences or get inferences for an entire dataset by using batch transform.
 
+
+Important Note:
+
+*  When using torch to load Models, it is recommended to use version torch>=2.6.0 and torchvision>=0.17.0
+
 Prepare a Training script
 =========================
 

src/sagemaker/_studio.py

@@ -65,7 +65,10 @@ def _find_config(working_dir=None):
         wd = Path(working_dir) if working_dir else Path.cwd()
 
         path = None
-        while path is None and not wd.match("/"):
+
+        # Get the root of the current working directory for both Windows and Unix-like systems
+        root = Path(wd.anchor)
+        while path is None and wd != root:
             candidate = wd / STUDIO_PROJECT_CONFIG
             if Path.exists(candidate):
                 path = candidate

src/sagemaker/estimator.py

@@ -2550,7 +2550,6 @@ def _get_train_args(cls, estimator, inputs, experiment_config):
                 raise ValueError(
                     "File URIs are supported in local mode only. Please use a S3 URI instead."
                 )
-
         config = _Job._load_config(inputs, estimator)
 
         current_hyperparameters = estimator.hyperparameters()

src/sagemaker/experiments/_metrics.py

@@ -197,8 +197,8 @@ def _send_metrics(self, metrics):
             response = self._metrics_client.batch_put_metrics(**request)
             errors = response["Errors"] if "Errors" in response else None
             if errors:
-                message = errors[0]["Message"]
-                raise Exception(f'{len(errors)} errors with message "{message}"')
+                error_code = errors[0]["Code"]
+                raise Exception(f'{len(errors)} errors with error code "{error_code}"')
 
     def _construct_batch_put_metrics_request(self, batch):
         """Creates dictionary object used as request to metrics service."""

src/sagemaker/huggingface/model.py

@@ -218,6 +218,7 @@ def deploy(
         container_startup_health_check_timeout=None,
         inference_recommendation_id=None,
         explainer_config=None,
+        update_endpoint: Optional[bool] = False,
         **kwargs,
     ):
         """Deploy this ``Model`` to an ``Endpoint`` and optionally return a ``Predictor``.
@@ -296,6 +297,11 @@ def deploy(
                 would like to deploy the model and endpoint with recommended parameters.
             explainer_config (sagemaker.explainer.ExplainerConfig): Specifies online explainability
                 configuration for use with Amazon SageMaker Clarify. (default: None)
+            update_endpoint (Optional[bool]):
+                Flag to update the model in an existing Amazon SageMaker endpoint.
+                If True, this will deploy a new EndpointConfig to an already existing endpoint
+                and delete resources corresponding to the previous EndpointConfig. Default: False
+                Note: Currently this is supported for single model endpoints
         Raises:
              ValueError: If arguments combination check failed in these circumstances:
                 - If no role is specified or
@@ -335,6 +341,7 @@ def deploy(
             container_startup_health_check_timeout=container_startup_health_check_timeout,
             inference_recommendation_id=inference_recommendation_id,
             explainer_config=explainer_config,
+            update_endpoint=update_endpoint,
             **kwargs,
         )
 

src/sagemaker/image_uri_config/pytorch.json

@@ -85,7 +85,8 @@
             "2.2": "2.2.0",
             "2.3": "2.3.0",
             "2.4": "2.4.0",
-            "2.5": "2.5.1"
+            "2.5": "2.5.1",
+            "2.6": "2.6.0"
         },
         "versions": {
             "0.4.0": {
@@ -1253,6 +1254,50 @@
                     "us-west-2": "763104351884"
                 },
                 "repository": "pytorch-inference"
+            },
+            "2.6.0": {
+                "py_versions": [
+                    "py312"
+                ],
+                "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-south-2": "772153158452",
+                    "ap-southeast-1": "763104351884",
+                    "ap-southeast-2": "763104351884",
+                    "ap-southeast-3": "907027046896",
+                    "ap-southeast-4": "457447274322",
+                    "ap-southeast-5": "550225433462",
+                    "ap-southeast-7": "590183813437",
+                    "ca-central-1": "763104351884",
+                    "ca-west-1": "204538143572",
+                    "cn-north-1": "727897471807",
+                    "cn-northwest-1": "727897471807",
+                    "eu-central-1": "763104351884",
+                    "eu-central-2": "380420809688",
+                    "eu-north-1": "763104351884",
+                    "eu-south-1": "692866216735",
+                    "eu-south-2": "503227376785",
+                    "eu-west-1": "763104351884",
+                    "eu-west-2": "763104351884",
+                    "eu-west-3": "763104351884",
+                    "il-central-1": "780543022126",
+                    "me-central-1": "914824155844",
+                    "me-south-1": "217643126080",
+                    "mx-central-1": "637423239942",
+                    "sa-east-1": "763104351884",
+                    "us-east-1": "763104351884",
+                    "us-east-2": "763104351884",
+                    "us-gov-east-1": "446045086412",
+                    "us-gov-west-1": "442386744353",
+                    "us-west-1": "763104351884",
+                    "us-west-2": "763104351884"
+                },
+                "repository": "pytorch-inference"
             }
         }
     },
@@ -1628,7 +1673,8 @@
             "2.2": "2.2.0",
             "2.3": "2.3.0",
             "2.4": "2.4.0",
-            "2.5": "2.5.1"
+            "2.5": "2.5.1",
+            "2.6": "2.6.0"
         },
         "versions": {
             "0.4.0": {
@@ -2801,6 +2847,50 @@
                     "us-west-2": "763104351884"
                 },
                 "repository": "pytorch-training"
+            },
+            "2.6.0": {
+                "py_versions": [
+                    "py312"
+                ],
+                "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-south-2": "772153158452",
+                    "ap-southeast-1": "763104351884",
+                    "ap-southeast-2": "763104351884",
+                    "ap-southeast-3": "907027046896",
+                    "ap-southeast-4": "457447274322",
+                    "ap-southeast-5": "550225433462",
+                    "ap-southeast-7": "590183813437",
+                    "ca-central-1": "763104351884",
+                    "ca-west-1": "204538143572",
+                    "cn-north-1": "727897471807",
+                    "cn-northwest-1": "727897471807",
+                    "eu-central-1": "763104351884",
+                    "eu-central-2": "380420809688",
+                    "eu-north-1": "763104351884",
+                    "eu-south-1": "692866216735",
+                    "eu-south-2": "503227376785",
+                    "eu-west-1": "763104351884",
+                    "eu-west-2": "763104351884",
+                    "eu-west-3": "763104351884",
+                    "il-central-1": "780543022126",
+                    "me-central-1": "914824155844",
+                    "me-south-1": "217643126080",
+                    "mx-central-1": "637423239942",
+                    "sa-east-1": "763104351884",
+                    "us-east-1": "763104351884",
+                    "us-east-2": "763104351884",
+                    "us-gov-east-1": "446045086412",
+                    "us-gov-west-1": "442386744353",
+                    "us-west-1": "763104351884",
+                    "us-west-2": "763104351884"
+                },
+                "repository": "pytorch-training"
             }
         }
     }

src/sagemaker/inputs.py

@@ -43,6 +43,8 @@ def __init__(
         attribute_names: Optional[List[Union[str, PipelineVariable]]] = None,
         target_attribute_name: Optional[Union[str, PipelineVariable]] = None,
         shuffle_config: Optional["ShuffleConfig"] = None,
+        hub_access_config: Optional[dict] = None,
+        model_access_config: Optional[dict] = None,
     ):
         r"""Create a definition for input data used by an SageMaker training job.
 
@@ -102,6 +104,13 @@ def __init__(
             shuffle_config (sagemaker.inputs.ShuffleConfig): If specified this configuration enables
                 shuffling on this channel. See the SageMaker API documentation for more info:
                 https://docs.aws.amazon.com/sagemaker/latest/dg/API_ShuffleConfig.html
+            hub_access_config (dict): Specify the HubAccessConfig of a
+                Model Reference for which a training job is being created for.
+            model_access_config (dict): For models that require a Model Access Config, specify True
+                or False for to indicate whether model terms of use have been accepted.
+                The `accept_eula` value must be explicitly defined as `True` in order to
+                accept the end-user license agreement (EULA) that some
+                models require. (Default: None).
         """
         self.config = {
             "DataSource": {"S3DataSource": {"S3DataType": s3_data_type, "S3Uri": s3_data}}
@@ -129,6 +138,27 @@ def __init__(
             self.config["TargetAttributeName"] = target_attribute_name
         if shuffle_config is not None:
             self.config["ShuffleConfig"] = {"Seed": shuffle_config.seed}
+        self.add_hub_access_config(hub_access_config)
+        self.add_model_access_config(model_access_config)
+
+    def add_hub_access_config(self, hub_access_config=None):
+        """Add Hub Access Config to the channel's configuration.
+
+        Args:
+            hub_access_config (dict): The HubAccessConfig to be added to the
+            channel's configuration.
+        """
+        if hub_access_config is not None:
+            self.config["DataSource"]["S3DataSource"]["HubAccessConfig"] = hub_access_config
+
+    def add_model_access_config(self, model_access_config=None):
+        """Add Model Access Config to the channel's configuration.
+
+        Args:
+            model_access_config (dict): Whether model terms of use have been accepted.
+        """
+        if model_access_config is not None:
+            self.config["DataSource"]["S3DataSource"]["ModelAccessConfig"] = model_access_config
 
 
 class ShuffleConfig(object):