Merge branch 'master-ci-py312-pr-test' into master-ci-py312-new

Author: benieric

CHANGELOG.md

@@ -1,5 +1,46 @@
 # Changelog
 
+## v2.243.0 (2025-03-27)
+
+### Features
+
+ * Enabled update_endpoint through model_builder
+
+### Bug Fixes and Other Changes
+
+ * Update for PT 2.5.1, SMP 2.8.0
+ * chore: move jumpstart region definitions to json file
+ * fix flaky clarify model monitor test
+ * fix flaky spark processor integ
+ * use temp file in unit tests
+ * Update transformers version
+ * Aligned disable_output_compression for @remote with Estimator
+ * Update Jinja version
+ * update image_uri_configs  03-26-2025 07:18:16 PST
+ * chore: fix integ tests to use latest version of model
+ * update image_uri_configs  03-25-2025 07:18:13 PST
+ * Skip tests failed due to deprecated instance type
+ * update image_uri_configs  03-21-2025 07:17:55 PST
+ * factor in set instance type when building JumpStart models in ModelBuilder.
+ * ADD Documentation to ReadtheDocs for Upgrading torch versions
+ * add new regions to JUMPSTART_LAUNCHED_REGIONS
+
+## 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

VERSION

@@ -1 +1 @@
-2.241.1.dev0
+2.243.1.dev0

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
 =========================
 

requirements/extras/test_requirements.txt

@@ -20,7 +20,7 @@ attrs>=23.1.0,<24
 fabric==2.6.0
 requests==2.32.2
 sagemaker-experiments==0.1.35
-Jinja2==3.1.4
+Jinja2==3.1.6
 pyvis==0.2.1
 pandas==1.4.4
 scikit-learn==1.3.0

src/sagemaker/fw_utils.py

@@ -155,6 +155,7 @@
     "2.3.0",
     "2.3.1",
     "2.4.1",
+    "2.5.1",
 ]
 
 TRAINIUM_SUPPORTED_DISTRIBUTION_STRATEGIES = ["torch_distributed"]

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,
         )