Merge branch 'master' into master-revert-xgboost

Author: benieric

CHANGELOG.md

@@ -1,5 +1,16 @@
 # Changelog
 
+## v2.237.1 (2024-12-12)
+
+### Bug Fixes and Other Changes
+
+ * chore: remove support for ecr spec fallbacks for jumpstart models
+ * Cloudpickle Revert
+ * Cloudpickle update
+ * Numpy update
+ * Protobuf update
+ * Update to fetch latest Cloudpickle version
+
 ## v2.237.0 (2024-12-05)
 
 ### Features

VERSION

@@ -1 +1 @@
-2.237.1.dev0
+2.237.2.dev0

pyproject.toml

@@ -33,13 +33,13 @@ classifiers = [
 dependencies = [
   "attrs>=23.1.0,<24",
   "boto3>=1.35.75,<2.0",
-  "cloudpickle>=2.2.1",
+  "cloudpickle==2.2.1",
   "docker",
   "fastapi",
   "google-pasta",
   "importlib-metadata>=1.4.0,<7.0",
   "jsonschema",
-  "numpy>=1.9.0,<3.0",
+  "numpy>=1.9.0,<2.0",
   "omegaconf>=2.2,<2.3",
   "packaging>=20.0",
   "pandas",

requirements/extras/test_requirements.txt

@@ -24,7 +24,7 @@ Jinja2==3.1.4
 pyvis==0.2.1
 pandas==1.4.4
 scikit-learn==1.3.0
-cloudpickle>=2.2.1
+cloudpickle==2.2.1
 jsonpickle<4.0.0
 PyYAML==6.0
 # TODO find workaround

src/sagemaker/image_uris.py

@@ -21,7 +21,7 @@
 from packaging.version import Version
 
 from sagemaker import utils
-from sagemaker.jumpstart.constants import DEFAULT_JUMPSTART_SAGEMAKER_SESSION
+from sagemaker.jumpstart.constants import DEFAULT_JUMPSTART_SAGEMAKER_SESSION, JUMPSTART_LOGGER
 from sagemaker.jumpstart.enums import JumpStartModelType
 from sagemaker.jumpstart.utils import is_jumpstart_model_input
 from sagemaker.spark import defaults
@@ -154,23 +154,27 @@ def retrieve(
             )
 
     if is_jumpstart_model_input(model_id, model_version):
+        if non_none_fields := {
+            key: value
+            for key, value in args.items()
+            if key in {"version", "framework", "container_version", "py_version"}
+            and value is not None
+        }:
+            JUMPSTART_LOGGER.info(
+                "Ignoring the following arguments when retrieving image uri "
+                "for JumpStart model id '%s': %s",
+                model_id,
+                str(non_none_fields),
+            )
         return artifacts._retrieve_image_uri(
-            model_id,
-            model_version,
-            image_scope,
-            hub_arn,
-            framework,
-            region,
-            version,
-            py_version,
-            instance_type,
-            accelerator_type,
-            container_version,
-            distribution,
-            base_framework_version,
-            training_compiler_config,
-            tolerate_vulnerable_model,
-            tolerate_deprecated_model,
+            model_id=model_id,
+            model_version=model_version,
+            image_scope=image_scope,
+            hub_arn=hub_arn,
+            region=region,
+            instance_type=instance_type,
+            tolerate_vulnerable_model=tolerate_vulnerable_model,
+            tolerate_deprecated_model=tolerate_deprecated_model,
             sagemaker_session=sagemaker_session,
             config_name=config_name,
             model_type=model_type,

src/sagemaker/jumpstart/artifacts/image_uris.py

@@ -14,14 +14,12 @@
 from __future__ import absolute_import
 
 from typing import Optional
-from sagemaker import image_uris
 from sagemaker.jumpstart.constants import (
     DEFAULT_JUMPSTART_SAGEMAKER_SESSION,
 )
 from sagemaker.jumpstart.enums import (
     JumpStartModelType,
     JumpStartScriptScope,
-    ModelFramework,
 )
 from sagemaker.jumpstart.utils import (
     get_region_fallback,
@@ -35,16 +33,8 @@ def _retrieve_image_uri(
     model_version: str,
     image_scope: str,
     hub_arn: Optional[str] = None,
-    framework: Optional[str] = None,
     region: Optional[str] = None,
-    version: Optional[str] = None,
-    py_version: Optional[str] = None,
     instance_type: Optional[str] = None,
-    accelerator_type: Optional[str] = None,
-    container_version: Optional[str] = None,
-    distribution: Optional[str] = None,
-    base_framework_version: Optional[str] = None,
-    training_compiler_config: Optional[str] = None,
     tolerate_vulnerable_model: bool = False,
     tolerate_deprecated_model: bool = False,
     sagemaker_session: Session = DEFAULT_JUMPSTART_SAGEMAKER_SESSION,
@@ -66,30 +56,11 @@ def _retrieve_image_uri(
         image_scope (str): The image type, i.e. what it is used for.
             Valid values: "training", "inference", "eia". If ``accelerator_type`` is set,
             ``image_scope`` is ignored.
-        framework (str): The name of the framework or algorithm.
         region (str): The AWS region. (Default: None).
-        version (str): The framework or algorithm version. This is required if there is
-            more than one supported version for the given framework or algorithm.
-            (Default: None).
-        py_version (str): The Python version. This is required if there is
-            more than one supported Python version for the given framework version.
         instance_type (str): The SageMaker instance type. For supported types, see
             https://aws.amazon.com/sagemaker/pricing/instance-types. This is required if
             there are different images for different processor types.
             (Default: None).
-        accelerator_type (str): Elastic Inference accelerator type. For more, see
-            https://docs.aws.amazon.com/sagemaker/latest/dg/ei.html.
-            (Default: None).
-        container_version (str): the version of docker image.
-            Ideally the value of parameter should be created inside the framework.
-            For custom use, see the list of supported container versions:
-            https://github.com/aws/deep-learning-containers/blob/master/available_images.md.
-            (Default: None).
-        distribution (dict): A dictionary with information on how to run distributed training.
-            (Default: None).
-        training_compiler_config (:class:`~sagemaker.training_compiler.TrainingCompilerConfig`):
-            A configuration class for the SageMaker Training Compiler.
-            (Default: None).
         tolerate_vulnerable_model (bool): True if vulnerable versions of model
             specifications should be tolerated (exception not raised). If False, raises an
             exception if the script used by this version of the model has dependencies with known
@@ -142,14 +113,12 @@ def _retrieve_image_uri(
             ecr_uri = model_specs.hosting_ecr_uri
             return ecr_uri
 
-        ecr_specs = model_specs.hosting_ecr_specs
-        if ecr_specs is None:
-            raise ValueError(
-                f"No inference ECR configuration found for JumpStart model ID '{model_id}' "
-                f"with {instance_type} instance type in {region}. "
-                "Please try another instance type or region."
-            )
-    elif image_scope == JumpStartScriptScope.TRAINING:
+        raise ValueError(
+            f"No inference ECR configuration found for JumpStart model ID '{model_id}' "
+            f"with {instance_type} instance type in {region}. "
+            "Please try another instance type or region."
+        )
+    if image_scope == JumpStartScriptScope.TRAINING:
         training_instance_type_variants = model_specs.training_instance_type_variants
         if training_instance_type_variants:
             image_uri = training_instance_type_variants.get_image_uri(
@@ -161,65 +130,10 @@ def _retrieve_image_uri(
             ecr_uri = model_specs.training_ecr_uri
             return ecr_uri
 
-        ecr_specs = model_specs.training_ecr_specs
-        if ecr_specs is None:
-            raise ValueError(
-                f"No training ECR configuration found for JumpStart model ID '{model_id}' "
-                f"with {instance_type} instance type in {region}. "
-                "Please try another instance type or region."
-            )
-    if framework is not None and framework != ecr_specs.framework:
-        raise ValueError(
-            f"Incorrect container framework '{framework}' for JumpStart model ID '{model_id}' "
-            f"and version '{model_version}'."
-        )
-
-    if version is not None and version != ecr_specs.framework_version:
-        raise ValueError(
-            f"Incorrect container framework version '{version}' for JumpStart model ID "
-            f"'{model_id}' and version '{model_version}'."
-        )
-
-    if py_version is not None and py_version != ecr_specs.py_version:
         raise ValueError(
-            f"Incorrect python version '{py_version}' for JumpStart model ID '{model_id}' "
-            f"and version '{model_version}'."
-        )
-
-    base_framework_version_override: Optional[str] = None
-    version_override: Optional[str] = None
-    if ecr_specs.framework == ModelFramework.HUGGINGFACE:
-        base_framework_version_override = ecr_specs.framework_version
-        version_override = ecr_specs.huggingface_transformers_version
-
-    if image_scope == JumpStartScriptScope.TRAINING:
-        return image_uris.get_training_image_uri(
-            region=region,
-            framework=ecr_specs.framework,
-            framework_version=version_override or ecr_specs.framework_version,
-            py_version=ecr_specs.py_version,
-            image_uri=None,
-            distribution=None,
-            compiler_config=None,
-            tensorflow_version=None,
-            pytorch_version=base_framework_version_override or base_framework_version,
-            instance_type=instance_type,
+            f"No training ECR configuration found for JumpStart model ID '{model_id}' "
+            f"with {instance_type} instance type in {region}. "
+            "Please try another instance type or region."
         )
-    if base_framework_version_override is not None:
-        base_framework_version_override = f"pytorch{base_framework_version_override}"
 
-    return image_uris.retrieve(
-        framework=ecr_specs.framework,
-        region=region,
-        version=version_override or ecr_specs.framework_version,
-        py_version=ecr_specs.py_version,
-        instance_type=instance_type,
-        hub_arn=hub_arn,
-        accelerator_type=accelerator_type,
-        image_scope=image_scope,
-        container_version=container_version,
-        distribution=distribution,
-        base_framework_version=base_framework_version_override or base_framework_version,
-        training_compiler_config=training_compiler_config,
-        config_name=config_name,
-    )
+    raise ValueError(f"Invalid scope: {image_scope}")

src/sagemaker/serve/utils/conda_in_process.yml

@@ -39,7 +39,7 @@ dependencies:
             - certifi==2022.12.7
             - charset-normalizer>=3.1.0
             - click>=8.1.3
-            - cloudpickle>=2.2.1
+            - cloudpickle==2.2.1
             - colorama>=0.4.4
             - contextlib2>=21.6.0
             - decorator>=5.1.1

src/sagemaker/serve/utils/in_process_requirements.txt

@@ -7,7 +7,7 @@ cachetools>=5.3.0
 certifi==2024.7.4
 charset-normalizer>=3.1.0
 click>=8.1.3
-cloudpickle>=2.2.1
+cloudpickle==2.2.1
 colorama>=0.4.4
 contextlib2>=21.6.0
 decorator>=5.1.1

tests/data/serve_resources/mlflow/pytorch/requirements.txt

@@ -1,7 +1,7 @@
 mlflow==2.13.2
 astunparse==1.6.3
 cffi==1.16.0
-cloudpickle>=2.2.1
+cloudpickle==2.2.1
 defusedxml==0.7.1
 dill==0.3.8
 gmpy2==2.1.2

tests/data/serve_resources/mlflow/tensorflow/requirements.txt

@@ -1,4 +1,4 @@
 mlflow==2.13.2
-cloudpickle>=2.2.1
+cloudpickle==2.2.1
 numpy==1.26.4
 tensorflow==2.16.1