sagemaker 2.147.0__tar.gz → 2.149.0__tar.gz

{sagemaker-2.147.0/src/sagemaker.egg-info → sagemaker-2.149.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sagemaker
-Version: 2.147.0
+Version: 2.149.0
 Summary: Open source library for training and deploying models on Amazon SageMaker.
 Home-page: https://github.com/aws/sagemaker-python-sdk/
 Author: Amazon Web Services
@@ -140,6 +140,8 @@ Description: .. image:: https://github.com/aws/sagemaker-python-sdk/raw/master/b
         1. AWS account credentials are available in the environment for the boto3 client to use.
         2. The AWS account has an IAM role named :code:`SageMakerRole`.
            It should have the AmazonSageMakerFullAccess policy attached as well as a policy with `the necessary permissions to use Elastic Inference <https://docs.aws.amazon.com/sagemaker/latest/dg/ei-setup.html>`__.
+        3. To run remote_function tests, dummy ecr repo should be created. It can be created by running -
+            :code:`aws ecr create-repository --repository-name remote-function-dummy-container`
         
         We recommend selectively running just those integration tests you'd like to run. You can filter by individual test function names with:
         

{sagemaker-2.147.0 → sagemaker-2.149.0}/README.rst

@@ -133,6 +133,8 @@ To run the integration tests, the following prerequisites must be met
 1. AWS account credentials are available in the environment for the boto3 client to use.
 2. The AWS account has an IAM role named :code:`SageMakerRole`.
    It should have the AmazonSageMakerFullAccess policy attached as well as a policy with `the necessary permissions to use Elastic Inference <https://docs.aws.amazon.com/sagemaker/latest/dg/ei-setup.html>`__.
+3. To run remote_function tests, dummy ecr repo should be created. It can be created by running -
+    :code:`aws ecr create-repository --repository-name remote-function-dummy-container`
 
 We recommend selectively running just those integration tests you'd like to run. You can filter by individual test function names with:
 

sagemaker-2.149.0/VERSION

@@ -0,0 +1 @@
+2.149.0

{sagemaker-2.147.0 → sagemaker-2.149.0}/requirements/extras/test_requirements.txt

@@ -21,3 +21,4 @@ sagemaker-experiments==0.1.35
 Jinja2==3.0.3
 pandas>=1.3.5,<1.5
 scikit-learn==1.0.2
+cloudpickle==2.2.1

{sagemaker-2.147.0 → sagemaker-2.149.0}/setup.py

@@ -49,6 +49,7 @@ def read_requirements(filename):
 required_packages = [
     "attrs>=20.3.0,<23",
     "boto3>=1.26.28,<2.0",
+    "cloudpickle==2.2.1",
     "google-pasta",
     "numpy>=1.9.0,<2.0",
     "protobuf>=3.1,<4.0",
@@ -62,6 +63,7 @@ required_packages = [
     "PyYAML==5.4.1",
     "jsonschema",
     "platformdirs",
+    "tblib==1.7.0",
 ]
 
 # Specific use case dependencies

{sagemaker-2.147.0 → sagemaker-2.149.0}/src/sagemaker/clarify.py

@@ -26,7 +26,7 @@ import re
 import tempfile
 from abc import ABC, abstractmethod
 from typing import List, Union, Dict, Optional, Any
-
+from enum import Enum
 from schema import Schema, And, Use, Or, Optional as SchemaOptional, Regex
 
 from sagemaker import image_uris, s3, utils
@@ -304,6 +304,16 @@ ANALYSIS_CONFIG_SCHEMA_V1_0 = Schema(
 )
 
 
+class DatasetType(Enum):
+    """Enum to store different dataset types supported in the Analysis config file"""
+
+    TEXTCSV = "text/csv"
+    JSONLINES = "application/jsonlines"
+    JSON = "application/json"
+    PARQUET = "application/x-parquet"
+    IMAGE = "application/x-image"
+
+
 class DataConfig:
     """Config object related to configurations of the input and output dataset."""
 
@@ -1451,7 +1461,7 @@ class SageMakerClarifyProcessor(Processor):
                 source=self._CLARIFY_OUTPUT,
                 destination=data_config.s3_output_path,
                 output_name="analysis_result",
-                s3_upload_mode="EndOfJob",
+                s3_upload_mode=ProcessingOutputHandler.get_s3_upload_mode(analysis_config),
             )
 
             return super().run(
@@ -2171,6 +2181,33 @@ def _upload_analysis_config(analysis_config_file, s3_output_path, sagemaker_sess
     )
 
 
+class ProcessingOutputHandler:
+    """Class to handle the parameters for SagemakerProcessor.Processingoutput"""
+
+    class S3UploadMode(Enum):
+        """Enum values for different uplaod modes to s3 bucket"""
+
+        CONTINUOUS = "Continuous"
+        ENDOFJOB = "EndOfJob"
+
+    @classmethod
+    def get_s3_upload_mode(cls, analysis_config: Dict[str, Any]) -> str:
+        """Fetches s3_upload mode based on the shap_config values
+
+        Args:
+            analysis_config (dict): dict Config following the analysis_config.json format
+
+        Returns:
+            The s3_upload_mode type for the processing output.
+        """
+        dataset_type = analysis_config["dataset_type"]
+        return (
+            ProcessingOutputHandler.S3UploadMode.CONTINUOUS.value
+            if dataset_type == DatasetType.IMAGE.value
+            else ProcessingOutputHandler.S3UploadMode.ENDOFJOB.value
+        )
+
+
 def _set(value, key, dictionary):
     """Sets dictionary[key] = value if value is not None."""
     if value is not None:

{sagemaker-2.147.0 → sagemaker-2.149.0}/src/sagemaker/config/config_schema.py

@@ -44,6 +44,17 @@ MONITORING_JOB_DEFINITION = "MonitoringJobDefinition"
 SAGEMAKER = "SageMaker"
 PYTHON_SDK = "PythonSDK"
 MODULES = "Modules"
+REMOTE_FUNCTION = "RemoteFunction"
+DEPENDENCIES = "Dependencies"
+PRE_EXECUTION_SCRIPT = "PreExecutionScript"
+PRE_EXECUTION_COMMANDS = "PreExecutionCommands"
+ENVIRONMENT_VARIABLES = "EnvironmentVariables"
+IMAGE_URI = "ImageUri"
+INCLUDE_LOCAL_WORKDIR = "IncludeLocalWorkDir"
+INSTANCE_TYPE = "InstanceType"
+S3_KMS_KEY_ID = "S3KmsKeyId"
+S3_ROOT_URI = "S3RootUri"
+JOB_CONDA_ENV = "JobCondaEnvironment"
 OFFLINE_STORE_CONFIG = "OfflineStoreConfig"
 ONLINE_STORE_CONFIG = "OnlineStoreConfig"
 S3_STORAGE_CONFIG = "S3StorageConfig"
@@ -221,6 +232,49 @@ MODEL_PACKAGE_VALIDATION_PROFILES_PATH = _simple_path(
     SAGEMAKER, MODEL_PACKAGE, VALIDATION_SPECIFICATION, VALIDATION_PROFILES
 )
 
+REMOTE_FUNCTION_DEPENDENCIES = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, DEPENDENCIES
+)
+REMOTE_FUNCTION_PRE_EXECUTION_COMMANDS = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, PRE_EXECUTION_COMMANDS
+)
+REMOTE_FUNCTION_PRE_EXECUTION_SCRIPT = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, PRE_EXECUTION_SCRIPT
+)
+REMOTE_FUNCTION_ENVIRONMENT_VARIABLES = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, ENVIRONMENT_VARIABLES
+)
+REMOTE_FUNCTION_IMAGE_URI = _simple_path(SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, IMAGE_URI)
+REMOTE_FUNCTION_INCLUDE_LOCAL_WORKDIR = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, INCLUDE_LOCAL_WORKDIR
+)
+REMOTE_FUNCTION_INSTANCE_TYPE = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, INSTANCE_TYPE
+)
+REMOTE_FUNCTION_JOB_CONDA_ENV = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, JOB_CONDA_ENV
+)
+REMOTE_FUNCTION_ROLE_ARN = _simple_path(SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, ROLE_ARN)
+REMOTE_FUNCTION_S3_KMS_KEY_ID = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, S3_KMS_KEY_ID
+)
+REMOTE_FUNCTION_S3_ROOT_URI = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, S3_ROOT_URI
+)
+REMOTE_FUNCTION_TAGS = _simple_path(SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, TAGS)
+REMOTE_FUNCTION_VOLUME_KMS_KEY_ID = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, VOLUME_KMS_KEY_ID
+)
+REMOTE_FUNCTION_VPC_CONFIG_SUBNETS = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, VPC_CONFIG, SUBNETS
+)
+REMOTE_FUNCTION_VPC_CONFIG_SECURITY_GROUP_IDS = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, VPC_CONFIG, SECURITY_GROUP_IDS
+)
+REMOTE_FUNCTION_ENABLE_INTER_CONTAINER_TRAFFIC_ENCRYPTION = _simple_path(
+    SAGEMAKER, PYTHON_SDK, MODULES, REMOTE_FUNCTION, ENABLE_INTER_CONTAINER_TRAFFIC_ENCRYPTION
+)
+
 # Paths for reference elsewhere in the SDK.
 # Names include the schema version since the paths could change with other schema versions
 MONITORING_SCHEDULE_INTER_CONTAINER_ENCRYPTION_PATH = _simple_path(
@@ -245,7 +299,6 @@ TRAINING_JOB_INTER_CONTAINER_ENCRYPTION_PATH = _simple_path(
     SAGEMAKER, TRAINING_JOB, ENABLE_INTER_CONTAINER_TRAFFIC_ENCRYPTION
 )
 
-
 SAGEMAKER_PYTHON_SDK_CONFIG_SCHEMA = {
     "$schema": "https://json-schema.org/draft/2020-12/schema",
     TYPE: OBJECT,
@@ -377,6 +430,23 @@ SAGEMAKER_PYTHON_SDK_CONFIG_SCHEMA = {
             "minItems": 0,
             "maxItems": 50,
         },
+        # Regex is taken from https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_CreateTrainingJob.html#sagemaker-CreateTrainingJob-request-Environment
+        "environmentVariables": {
+            TYPE: OBJECT,
+            ADDITIONAL_PROPERTIES: False,
+            PATTERN_PROPERTIES: {
+                r"([a-zA-Z_][a-zA-Z0-9_]*){1,512}": {
+                    TYPE: "string",
+                    "pattern": r"[\S\s]*",
+                    "maxLength": 512,
+                }
+            },
+            "maxProperties": 48,
+        },
+        # Regex is taken from https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_S3DataSource.html#sagemaker-Type-S3DataSource-S3Uri
+        "s3Uri": {TYPE: "string", "pattern": "^(https|s3)://([^/]+)/?(.*)$", "maxLength": 1024},
+        # Regex is taken from https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_AlgorithmSpecification.html#sagemaker-Type-AlgorithmSpecification-ContainerEntrypoint
+        "preExecutionCommand": {TYPE: "string", "pattern": r".*"},
     },
     PROPERTIES: {
         SCHEMA_VERSION: {
@@ -406,6 +476,36 @@ SAGEMAKER_PYTHON_SDK_CONFIG_SCHEMA = {
                             # Any SageMaker Python SDK specific configuration will be added here.
                             TYPE: OBJECT,
                             ADDITIONAL_PROPERTIES: False,
+                            PROPERTIES: {
+                                REMOTE_FUNCTION: {
+                                    TYPE: OBJECT,
+                                    ADDITIONAL_PROPERTIES: False,
+                                    PROPERTIES: {
+                                        DEPENDENCIES: {TYPE: "string"},
+                                        PRE_EXECUTION_COMMANDS: {
+                                            TYPE: "array",
+                                            "items": {"$ref": "#/definitions/preExecutionCommand"},
+                                        },
+                                        PRE_EXECUTION_SCRIPT: {TYPE: "string"},
+                                        ENABLE_INTER_CONTAINER_TRAFFIC_ENCRYPTION: {
+                                            TYPE: "boolean"
+                                        },
+                                        ENVIRONMENT_VARIABLES: {
+                                            "$ref": "#/definitions/environmentVariables"
+                                        },
+                                        IMAGE_URI: {TYPE: "string"},
+                                        INCLUDE_LOCAL_WORKDIR: {TYPE: "boolean"},
+                                        INSTANCE_TYPE: {TYPE: "string"},
+                                        JOB_CONDA_ENV: {TYPE: "string"},
+                                        ROLE_ARN: {"$ref": "#/definitions/roleArn"},
+                                        S3_KMS_KEY_ID: {"$ref": "#/definitions/kmsKeyId"},
+                                        S3_ROOT_URI: {"$ref": "#/definitions/s3Uri"},
+                                        TAGS: {"$ref": "#/definitions/tags"},
+                                        VOLUME_KMS_KEY_ID: {"$ref": "#/definitions/kmsKeyId"},
+                                        VPC_CONFIG: {"$ref": "#/definitions/vpcConfig"},
+                                    },
+                                }
+                            },
                         }
                     },
                 },

{sagemaker-2.147.0 → sagemaker-2.149.0}/src/sagemaker/experiments/run.py

@@ -715,6 +715,14 @@ class Run(object):
 
         self.close()
 
+    def __getstate__(self):
+        """Overriding this method to prevent instance of Run from being pickled.
+
+        Raise:
+            NotImplementedError: If attempting to pickle this instance.
+        """
+        raise NotImplementedError("Instance of Run type is not allowed to be pickled.")
+
 
 def load_run(
     run_name: Optional[str] = None,
@@ -787,36 +795,38 @@ def load_run(
     Returns:
         Run: The loaded Run object.
     """
-    sagemaker_session = sagemaker_session or _utils.default_session()
     environment = _RunEnvironment.load()
 
     verify_load_input_names(run_name=run_name, experiment_name=experiment_name)
 
-    if run_name or environment:
-        if run_name:
-            logger.warning(
-                "run_name is explicitly supplied in load_run, "
-                "which will be prioritized to load the Run object. "
-                "In other words, the run name in the experiment config, fetched from the "
-                "job environment or the current run context, will be ignored."
-            )
-        else:
-            exp_config = get_tc_and_exp_config_from_job_env(
-                environment=environment, sagemaker_session=sagemaker_session
-            )
-            run_name = Run._extract_run_name_from_tc_name(
-                trial_component_name=exp_config[RUN_NAME],
-                experiment_name=exp_config[EXPERIMENT_NAME],
-            )
-            experiment_name = exp_config[EXPERIMENT_NAME]
-
+    if run_name:
+        logger.warning(
+            "run_name is explicitly supplied in load_run, "
+            "which will be prioritized to load the Run object. "
+            "In other words, the run name in the experiment config, fetched from the "
+            "job environment or the current run context, will be ignored."
+        )
         run_instance = Run(
             experiment_name=experiment_name,
             run_name=run_name,
-            sagemaker_session=sagemaker_session,
+            sagemaker_session=sagemaker_session or _utils.default_session(),
         )
     elif _RunContext.get_current_run():
         run_instance = _RunContext.get_current_run()
+    elif environment:
+        exp_config = get_tc_and_exp_config_from_job_env(
+            environment=environment, sagemaker_session=sagemaker_session or _utils.default_session()
+        )
+        run_name = Run._extract_run_name_from_tc_name(
+            trial_component_name=exp_config[RUN_NAME],
+            experiment_name=exp_config[EXPERIMENT_NAME],
+        )
+        experiment_name = exp_config[EXPERIMENT_NAME]
+        run_instance = Run(
+            experiment_name=experiment_name,
+            run_name=run_name,
+            sagemaker_session=sagemaker_session or _utils.default_session(),
+        )
     else:
         raise RuntimeError(
             "Failed to load a Run object. "

{sagemaker-2.147.0 → sagemaker-2.149.0}/src/sagemaker/fw_utils.py

@@ -135,6 +135,7 @@ SM_DATAPARALLEL_SUPPORTED_FRAMEWORK_VERSIONS = {
         "1.12.0",
         "1.12.1",
         "1.13.1",
+        "2.0.0",
     ],
 }
 
@@ -148,10 +149,11 @@ PYTORCHDDP_SUPPORTED_FRAMEWORK_VERSIONS = [
     "1.12.0",
     "1.12.1",
     "1.13.1",
+    "2.0.0",
 ]
 
 
-TORCH_DISTRIBUTED_GPU_SUPPORTED_FRAMEWORK_VERSIONS = ["1.13.1"]
+TORCH_DISTRIBUTED_GPU_SUPPORTED_FRAMEWORK_VERSIONS = ["1.13.1", "2.0.0"]
 
 TRAINIUM_SUPPORTED_DISTRIBUTION_STRATEGIES = ["torch_distributed"]
 TRAINIUM_SUPPORTED_TORCH_DISTRIBUTED_FRAMEWORK_VERSIONS = [
@@ -161,6 +163,7 @@ TRAINIUM_SUPPORTED_TORCH_DISTRIBUTED_FRAMEWORK_VERSIONS = [
     "1.12.0",
     "1.12.1",
     "1.13.1",
+    "2.0.0",
 ]
 
 SMDISTRIBUTED_SUPPORTED_STRATEGIES = ["dataparallel", "modelparallel"]

{sagemaker-2.147.0 → sagemaker-2.149.0}/src/sagemaker/huggingface/estimator.py

@@ -17,12 +17,10 @@ import logging
 import re
 from typing import Optional, Union, Dict
 
-from sagemaker.deprecations import renamed_kwargs
 from sagemaker.estimator import Framework, EstimatorBase
 from sagemaker.fw_utils import (
     framework_name_from_image,
-    warn_if_parameter_server_with_multi_gpu,
-    validate_smdistributed,
+    validate_distribution,
 )
 from sagemaker.huggingface.model import HuggingFaceModel
 from sagemaker.vpc_utils import VPC_CONFIG_DEFAULT
@@ -37,6 +35,9 @@ class HuggingFace(Framework):
     """Handle training of custom HuggingFace code."""
 
     _framework_name = "huggingface"
+    LAUNCH_PYTORCH_DDP_ENV_NAME = "sagemaker_pytorch_ddp_enabled"
+    LAUNCH_TORCH_DISTRIBUTED_ENV_NAME = "sagemaker_torch_distributed_enabled"
+    INSTANCE_TYPE_ENV_NAME = "sagemaker_instance_type"
 
     def __init__(
         self,
@@ -142,6 +143,36 @@ class HuggingFace(Framework):
                         }
                     }
 
+                **To enable PyTorch DDP:**
+
+                    .. code:: python
+
+                        {
+                            "pytorchddp": {
+                                "enabled": True
+                            }
+                        }
+
+                    To learn more, see `Distributed PyTorch Training
+                    <https://sagemaker.readthedocs.io/en/stable/frameworks/pytorch/using_pytorch.html#distributed-pytorch-training>`_.
+
+                **To enable Torch Distributed:**
+
+                    This is available for general distributed training on
+                    GPU instances from PyTorch v1.13.1 and later.
+
+                    .. code:: python
+
+                        {
+                            "torch_distributed": {
+                                "enabled": True
+                            }
+                        }
+
+                    This option also supports distributed training on Trn1.
+                    To learn more, see `Distributed PyTorch Training on Trainium
+                    <https://sagemaker.readthedocs.io/en/stable/frameworks/pytorch/using_pytorch.html#distributed-pytorch-training-on-trainium>`_.
+
                 To enable distributed training with
                 `SageMaker Training Compiler <https://docs.aws.amazon.com/sagemaker/latest/dg/training-compiler.html>`_
                 for Hugging Face Transformers with PyTorch:
@@ -182,29 +213,6 @@ class HuggingFace(Framework):
 
         self._validate_args(image_uri=image_uri)
 
-        instance_type = renamed_kwargs(
-            "train_instance_type", "instance_type", kwargs.get("instance_type"), kwargs
-        )
-
-        base_framework_name = "tensorflow" if tensorflow_version is not None else "pytorch"
-        base_framework_version = (
-            tensorflow_version if tensorflow_version is not None else pytorch_version
-        )
-
-        if distribution is not None:
-            validate_smdistributed(
-                instance_type=instance_type,
-                framework_name=base_framework_name,
-                framework_version=base_framework_version,
-                py_version=self.py_version,
-                distribution=distribution,
-                image_uri=image_uri,
-            )
-
-            warn_if_parameter_server_with_multi_gpu(
-                training_instance_type=instance_type, distribution=distribution
-            )
-
         if "enable_sagemaker_metrics" not in kwargs:
             kwargs["enable_sagemaker_metrics"] = True
 
@@ -214,6 +222,25 @@ class HuggingFace(Framework):
             entry_point, source_dir, hyperparameters, image_uri=image_uri, **kwargs
         )
 
+        if "entry_point" not in kwargs:
+            kwargs["entry_point"] = entry_point
+
+        self.base_framework_name = "tensorflow" if tensorflow_version is not None else "pytorch"
+        self.base_framework_version = (
+            tensorflow_version if tensorflow_version is not None else pytorch_version
+        )
+
+        if distribution is not None:
+            distribution = validate_distribution(
+                distribution,
+                self.instance_groups,
+                self.base_framework_name,
+                self.base_framework_version,
+                py_version,
+                image_uri,
+                kwargs,
+            )
+
         self.distribution = distribution or {}
 
         if compiler_config is not None:
@@ -267,14 +294,44 @@ class HuggingFace(Framework):
                 "transformers_version, tensorflow_version and pytorch_version."
             )
 
+    def _huggingface_distribution_configuration(self, distribution):
+        """Returns a dict of distribution config for Hugging Face training
+
+        Args:
+            distribution (dict): A dictionary with information on how to run distributed training.
+        Returns:
+            dict containing Pytorch DDP config
+        """
+        distribution_config = {}
+        pytorch_ddp_enabled = False
+        torch_distributed_enabled = False
+
+        if "pytorchddp" in distribution:
+            pytorch_ddp_enabled = distribution.get("pytorchddp").get("enabled", False)
+        elif "torch_distributed" in distribution:
+            torch_distributed_enabled = distribution.get("torch_distributed").get("enabled", False)
+
+        if pytorch_ddp_enabled:
+            distribution_config[self.LAUNCH_PYTORCH_DDP_ENV_NAME] = pytorch_ddp_enabled
+            if self.instance_type is not None:
+                distribution_config[self.INSTANCE_TYPE_ENV_NAME] = self.instance_type
+        elif torch_distributed_enabled:
+            distribution_config[self.LAUNCH_TORCH_DISTRIBUTED_ENV_NAME] = torch_distributed_enabled
+            if self.instance_type is not None:
+                distribution_config[self.INSTANCE_TYPE_ENV_NAME] = self.instance_type
+        else:
+            distribution_config = self._distribution_configuration(distribution=distribution)
+
+        return distribution_config
+
     def hyperparameters(self):
         """Return hyperparameters used by your custom PyTorch code during model training."""
         hyperparameters = super(HuggingFace, self).hyperparameters()
-        distributed_training_hyperparameters = self._distribution_configuration(
+        additional_hyperparameters = self._huggingface_distribution_configuration(
             distribution=self.distribution
         )
         hyperparameters.update(
-            EstimatorBase._json_encode_hyperparameters(distributed_training_hyperparameters)
+            EstimatorBase._json_encode_hyperparameters(additional_hyperparameters)
         )
 
         if self.compiler_config: