autogluon.cloud 0.5.0__tar.gz → 0.6.0__tar.gz

{autogluon_cloud-0.5.0/src/autogluon.cloud.egg-info → autogluon_cloud-0.6.0}/PKG-INFO

@@ -1,11 +1,10 @@
 Metadata-Version: 2.4
 Name: autogluon.cloud
-Version: 0.5.0
-Summary: Train and deploy AutoGluon backed models on the cloud
-Home-page: https://github.com/autogluon/autogluon-cloud
+Version: 0.6.0
+Summary: Train and deploy ML models in the cloud
 Author: AutoGluon Community
-License: Apache-2.0
-Project-URL: Documentation, https://auto.gluon.ai
+License-Expression: Apache-2.0
+Project-URL: Documentation, https://auto.gluon.ai/cloud
 Project-URL: Bug Reports, https://github.com/autogluon/autogluon-cloud/issues
 Project-URL: Source, https://github.com/autogluon/autogluon-cloud/
 Project-URL: Contribute!, https://github.com/autogluon/autogluon-cloud/blob/master/CONTRIBUTING.md
@@ -17,7 +16,6 @@ Classifier: Intended Audience :: Customer Service
 Classifier: Intended Audience :: Financial and Insurance Industry
 Classifier: Intended Audience :: Healthcare Industry
 Classifier: Intended Audience :: Telecommunications Industry
-Classifier: License :: OSI Approved :: Apache Software License
 Classifier: Operating System :: MacOS
 Classifier: Operating System :: Microsoft :: Windows
 Classifier: Operating System :: POSIX
@@ -31,14 +29,14 @@ Classifier: Topic :: Software Development
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Classifier: Topic :: Scientific/Engineering :: Information Analysis
 Classifier: Topic :: Scientific/Engineering :: Image Recognition
-Requires-Python: >=3.10, <3.14
+Requires-Python: <3.14,>=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
 License-File: NOTICE
-Requires-Dist: autogluon.common<1.6,>=0.7
-Requires-Dist: boto3<2,>=1.10
+Requires-Dist: autogluon.common<1.7,>=0.7
+Requires-Dist: boto3<2,>=1.37.23
 Requires-Dist: packaging<27,>=23.0
-Requires-Dist: sagemaker<3,>=2.126.0
+Requires-Dist: sagemaker<3,>=2.240.0
 Requires-Dist: pyarrow<25,>=19.0.1
 Requires-Dist: PyYAML~=6.0
 Requires-Dist: Pillow<13,>=10.2
@@ -47,28 +45,15 @@ Requires-Dist: typing_extensions<5,>=4.0
 Requires-Dist: click<9,>=8.0
 Requires-Dist: rich<15,>=13.0
 Provides-Extra: ray
-Requires-Dist: ray[default]<2.56,>=2.10.0; extra == "ray"
+Requires-Dist: ray[default]<2.59,>=2.10.0; extra == "ray"
 Provides-Extra: all
-Requires-Dist: autogluon<1.6,>=0.7; extra == "all"
-Requires-Dist: ray[default]<2.56,>=2.10.0; extra == "all"
+Requires-Dist: autogluon<1.7,>=0.7; extra == "all"
+Requires-Dist: autogluon.cloud[ray]; extra == "all"
 Provides-Extra: tests
-Requires-Dist: moto; extra == "tests"
 Requires-Dist: pytest; extra == "tests"
+Requires-Dist: moto; extra == "tests"
 Requires-Dist: autogluon.common>=0.7; extra == "tests"
-Requires-Dist: pytest-cov; extra == "tests"
-Requires-Dist: tox; extra == "tests"
-Dynamic: author
-Dynamic: classifier
-Dynamic: description
-Dynamic: description-content-type
-Dynamic: home-page
-Dynamic: license
 Dynamic: license-file
-Dynamic: project-url
-Dynamic: provides-extra
-Dynamic: requires-dist
-Dynamic: requires-python
-Dynamic: summary
 
 
 

autogluon_cloud-0.6.0/pyproject.toml

@@ -0,0 +1,109 @@
+[build-system]
+requires = ["setuptools>=77", "wheel"]  # >=77 for PEP 639 SPDX license expression support
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "autogluon.cloud"
+description = "Train and deploy ML models in the cloud"
+readme = "README.md"
+requires-python = ">=3.10,<3.14"
+license = "Apache-2.0"
+license-files = ["LICENSE", "NOTICE"]
+authors = [{ name = "AutoGluon Community" }]
+classifiers = [
+    "Development Status :: 4 - Beta",
+    "Intended Audience :: Education",
+    "Intended Audience :: Developers",
+    "Intended Audience :: Science/Research",
+    "Intended Audience :: Customer Service",
+    "Intended Audience :: Financial and Insurance Industry",
+    "Intended Audience :: Healthcare Industry",
+    "Intended Audience :: Telecommunications Industry",
+    "Operating System :: MacOS",
+    "Operating System :: Microsoft :: Windows",
+    "Operating System :: POSIX",
+    "Operating System :: Unix",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.10",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Topic :: Software Development",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+    "Topic :: Scientific/Engineering :: Information Analysis",
+    "Topic :: Scientific/Engineering :: Image Recognition",
+]
+# version stays dynamic and is owned by setup.py (so the RELEASE=1 -> clean
+# version, else .dev0 logic keeps working).
+dynamic = ["version"]
+dependencies = [
+    # common module provides utils with stable api across minor version
+    "autogluon.common>=0.7,<1.7",
+    # <2 because unlikely to introduce breaking changes in minor releases.
+    # TransformAmiVersion was added to the SageMaker service model in botocore 1.37.23.
+    "boto3>=1.37.23,<2",
+    "packaging>=23.0,<27",
+    # Supports forwarding inference_ami_version from Model.deploy.
+    "sagemaker>=2.240.0,<3",
+    "pyarrow>=19.0.1,<25",  # lower bound to avoid https://github.com/apache/arrow/issues/45283
+    "PyYAML~=6.0",
+    "Pillow>=10.2,<13",
+    "huggingface_hub>=0.20,<2",
+    "typing_extensions>=4.0,<5",
+    # CLI dependencies (autogluon-cloud command)
+    "click>=8.0,<9",
+    "rich>=13.0,<15",
+]
+
+[project.optional-dependencies]
+ray = ["ray[default]>=2.10.0,<2.59"]
+all = [  # To allow user to pass ag objects
+    "autogluon>=0.7,<1.7",
+    "autogluon.cloud[ray]",
+]
+tests = [
+    "pytest",
+    "moto",
+    "autogluon.common>=0.7",
+]
+
+[project.urls]
+Documentation = "https://auto.gluon.ai/cloud"
+"Bug Reports" = "https://github.com/autogluon/autogluon-cloud/issues"
+Source = "https://github.com/autogluon/autogluon-cloud/"
+"Contribute!" = "https://github.com/autogluon/autogluon-cloud/blob/master/CONTRIBUTING.md"
+
+[project.scripts]
+autogluon-cloud = "autogluon.cloud.cli:main"
+
+[tool.setuptools]
+zip-safe = true
+include-package-data = true
+
+[tool.setuptools.packages.find]
+where = ["src"]
+namespaces = true
+
+[tool.setuptools.package-data]
+"autogluon.cloud" = [
+    "default_cluster_configs/*.yaml",
+    "utils/autogluon_dlc.json",
+    "templates/*.yaml",
+]
+
+[tool.ruff]
+line-length = 119
+target-version = "py310"
+
+[tool.ruff.lint]
+select = ["E", "F", "I", "W"]
+ignore = ["E501"]
+per-file-ignores = {"*/__init__.py" = ["F401"]}
+
+[tool.ruff.lint.isort]
+known-first-party = ["autogluon"]
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+addopts = "--strict-markers"
+xfail_strict = true

autogluon_cloud-0.6.0/setup.py

@@ -0,0 +1,33 @@
+import os
+
+from setuptools import setup
+
+AUTOGLUON = "autogluon"
+CLOUD = "cloud"
+
+
+def create_version_file(*, version):
+    print("-- Building version " + version)
+    version_path = os.path.join("src", AUTOGLUON, CLOUD, "version.py")
+    with open(version_path, "w") as f:
+        f.write(f'"""This is the {AUTOGLUON}.{CLOUD} version file."""\n')
+        f.write("__version__ = '{}'\n".format(version))
+
+
+def update_version(version):
+    """
+    To release a new stable version on PyPi, tag the release on github; the Github CI publishes it to PyPi
+    (see .github/workflows/pypi_release.yml, which sets RELEASE=1). Non-release builds get a `.dev0` suffix.
+    """
+    if not os.getenv("RELEASE"):
+        version += ".dev0"
+    return version
+
+
+version = "0.6.0"
+version = update_version(version)
+
+if __name__ == "__main__":
+    create_version_file(version=version)
+    # Everything except the dynamic version lives in pyproject.toml.
+    setup(version=version)

{autogluon_cloud-0.5.0 → autogluon_cloud-0.6.0}/src/autogluon/cloud/__init__.py

@@ -1,5 +1,7 @@
 import logging
+import os
 
+os.environ.setdefault("SAGEMAKER_SUPPRESS_V2_WARNING", "1")
 import sagemaker
 from packaging.version import Version
 
@@ -12,8 +14,9 @@ if Version(sagemaker.__version__) >= Version("3.0"):
 from autogluon.common.utils.log_utils import _add_stream_handler
 
 from .cloud_setup import bootstrap, register, status, teardown
+from .endpoint.tabular_endpoint import TabularEndpoint
 from .endpoint.timeseries_endpoint import TimeSeriesEndpoint
-from .model.foundation_model import TimeSeriesFoundationModel
+from .model.foundation_model import TabularFoundationModel, TimeSeriesFoundationModel
 from .predictor import MultiModalCloudPredictor, TabularCloudPredictor, TimeSeriesCloudPredictor
 
 _add_stream_handler()
@@ -22,6 +25,8 @@ logging.getLogger(__name__).setLevel(logging.INFO)
 __all__ = [
     "MultiModalCloudPredictor",
     "TabularCloudPredictor",
+    "TabularEndpoint",
+    "TabularFoundationModel",
     "TimeSeriesCloudPredictor",
     "TimeSeriesEndpoint",
     "TimeSeriesFoundationModel",

{autogluon_cloud-0.5.0 → autogluon_cloud-0.6.0}/src/autogluon/cloud/backend/backend.py

@@ -1,7 +1,7 @@
 from __future__ import annotations
 
+import json
 import os
-import pickle
 from abc import ABC, abstractmethod
 from typing import Any, Dict, List, Optional, Union
 
@@ -10,6 +10,34 @@ import pandas as pd
 from ..endpoint.endpoint import Endpoint
 
 
+def dumps_ag_args(config: Dict[str, Any]) -> str:
+    """Serialize the remote-training config to JSON, raising a user-facing error on failure.
+
+    The config carries the user's predictor init/fit arguments. Some objects (search spaces,
+    custom metric objects, classes/callables in ``hyperparameters``) are not JSON-serializable;
+    when that happens we pinpoint the offending argument so the error names what the user passed.
+    """
+    try:
+        return json.dumps(config)
+    except TypeError as e:
+        original_error = e
+    for group in ("predictor_init_args", "predictor_fit_args"):
+        for key, value in (config.get(group) or {}).items():
+            try:
+                json.dumps(value)
+            except TypeError as e:
+                raise TypeError(
+                    f"The value passed for `{key}` is not JSON-serializable. This can happen with "
+                    f"search spaces (Real/Categorical/Int), custom metric objects, or classes/callables "
+                    f"in `hyperparameters`. Please pass these as plain values (e.g. metric names as strings). "
+                    f"Original error: {e}"
+                ) from e
+    # Culprit is outside the known arg groups; re-raise the original error with generic guidance.
+    raise TypeError(
+        f"The provided arguments are not JSON-serializable. Original error: {original_error}"
+    ) from original_error
+
+
 class Backend(ABC):
     name = "backend"
 
@@ -27,12 +55,18 @@ class Backend(ABC):
         return self._cloud_output_path
 
     def initialize(
-        self, local_output_path: str, predictor_type: str, cloud_output_path: Optional[str] = None, **kwargs
+        self,
+        local_output_path: str,
+        predictor_type: str,
+        cloud_output_path: Optional[str] = None,
+        resource_prefix: Optional[str] = None,
+        **kwargs,
     ) -> None:
         """Initialize the backend."""
         self.local_output_path = local_output_path
         self._cloud_output_path = cloud_output_path
         self.predictor_type = predictor_type
+        self.resource_prefix = resource_prefix or f"ag-cloud-{predictor_type}"
         self.original_features = None
         self.endpoint: Optional[Endpoint] = None
 
@@ -83,18 +117,19 @@ class Backend(ABC):
     def prepare_args(self, path: str, **kwargs):
         """
         prepare parameter args required to be passed to remote AG, i.e. init args and fit args
-        The args will be saved as a pickle object
+        The args will be saved as a JSON file
 
         Parameters
         ----------
         path: str
-            Path to save the pickle file
+            Path to save the JSON file
         """
         assert self.predictor_type is not None
         config = self._construct_ag_args(**kwargs)
+        payload = dumps_ag_args(config)
         os.makedirs(os.path.dirname(path), exist_ok=True)
-        with open(path, "wb") as f:
-            pickle.dump(config, f)
+        with open(path, "w") as f:
+            f.write(payload)
 
     def _construct_ag_args(**kwargs):
         raise NotImplementedError

{autogluon_cloud-0.5.0 → autogluon_cloud-0.6.0}/src/autogluon/cloud/backend/backend_factory.py

@@ -1,3 +1,5 @@
+import logging
+
 from .backend import Backend
 from .constant import RAY_AWS, TABULAR_RAY_AWS
 from .multimodal_sagemaker_backend import MultiModalSagemakerBackend
@@ -5,6 +7,8 @@ from .sagemaker_backend import SagemakerBackend
 from .tabular_sagemaker_backend import TabularSagemakerBackend
 from .timeseries_sagemaker_backend import TimeSeriesSagemakerBackend
 
+logger = logging.getLogger(__name__)
+
 
 class BackendFactory:
     _SAGEMAKER_BACKENDS = {
@@ -20,6 +24,10 @@ class BackendFactory:
         if backend in BackendFactory._SAGEMAKER_BACKENDS:
             return BackendFactory._SAGEMAKER_BACKENDS[backend]
         if backend in BackendFactory._RAY_BACKEND_NAMES:
+            logger.warning(
+                f"The '{backend}' backend is experimental and not fully supported. Use at your own risk; "
+                "for production workloads prefer the 'sagemaker' backend."
+            )
             try:
                 from .ray_aws_backend import RayAWSBackend, TabularRayAWSBackend
             except ImportError as e:

{autogluon_cloud-0.5.0 → autogluon_cloud-0.6.0}/src/autogluon/cloud/backend/ray_backend.py

@@ -17,7 +17,6 @@ from ..data import FormatConverterFactory
 from ..endpoint.endpoint import Endpoint
 from ..job.ray_job import RayFitJob
 from ..scripts import ScriptManager
-from ..utils.constants import CLOUD_RESOURCE_PREFIX
 from ..utils.dlc_utils import parse_framework_version, retrieve_image_uri
 from ..utils.ec2 import get_latest_ami
 from ..utils.iam import get_instance_profile_arn
@@ -143,7 +142,7 @@ class RayBackend(Backend):
             If `custom_image_uri` is set, this argument will be ignored.
         job_name: str, default = None
             Name of the launched training job.
-            If None, CloudPredictor will create one with prefix ag-cloudpredictor
+            If None, AutoGluon Cloud creates one with a predictor-specific prefix.
         instance_type: str, default = 'ml.m5.2xlarge'
             Instance type the predictor will be trained on with SageMaker.
         instance_count: Union[int, str], default = "auto",
@@ -211,7 +210,7 @@ class RayBackend(Backend):
         image_uri = self._get_image_uri(
             framework_version=framework_version, custom_image_uri=custom_image_uri, instance_type=instance_type
         )
-        ag_args_path = os.path.join(self.local_output_path, "job", "ag_args.pkl")
+        ag_args_path = os.path.join(self.local_output_path, "job", "ag_args.json")
         self.prepare_args(
             path=ag_args_path, predictor_init_args=predictor_init_args, predictor_fit_args=predictor_fit_args
         )
@@ -257,7 +256,7 @@ class RayBackend(Backend):
             cluster_manager.setup_connection()
             time.sleep(10)  # waiting for connection to setup
             if job_name is None:
-                job_name = CLOUD_RESOURCE_PREFIX + "-" + get_utc_timestamp_now()
+                job_name = self.resource_prefix + "-" + get_utc_timestamp_now()
             job = RayFitJob(output_path=self.cloud_output_path + "/model")
             self._fit_job = job
 

{autogluon_cloud-0.5.0 → autogluon_cloud-0.6.0}/src/autogluon/cloud/backend/sagemaker_backend.py

@@ -2,7 +2,6 @@ import copy
 import json
 import logging
 import os
-import pickle
 import shutil
 import tarfile
 import tempfile
@@ -28,8 +27,8 @@ from ..utils.ag_sagemaker import (
     AutoGluonRepackInferenceModel,
 )
 from ..utils.aws_utils import resolve_execution_role, setup_sagemaker_session
-from ..utils.constants import CLOUD_RESOURCE_PREFIX, VALID_ACCEPT
-from ..utils.dlc_utils import parse_framework_version
+from ..utils.constants import VALID_ACCEPT
+from ..utils.dlc_utils import infer_sagemaker_ami_version, parse_framework_version
 from ..utils.misc import MostRecentInsertedOrderedDict
 from ..utils.serializers import AutoGluonSerializationWrapper
 from ..utils.tag_utils import build_tags
@@ -208,7 +207,7 @@ class SagemakerBackend(Backend):
             If `custom_image_uri` is set, this argument will be ignored.
         job_name: str, default = None
             Name of the launched training job.
-            If None, CloudPredictor will create one with prefix ag-cloudpredictor
+            If None, AutoGluon Cloud creates one with a predictor- or model-specific prefix.
         instance_type: str, default = 'ml.m5.2xlarge'
             Instance type the predictor will be trained on with SageMaker.
         instance_count: int, default = 1
@@ -224,13 +223,14 @@ class SagemakerBackend(Backend):
             Use `get_fit_job_status` to get job status.
         autogluon_sagemaker_estimator_kwargs: dict, default = dict()
             Any extra arguments needed to initialize AutoGluonSagemakerEstimator
-            Please refer to https://sagemaker.readthedocs.io/en/stable/api/training/estimators.html#sagemaker.estimator.Framework for all options
+            Please refer to https://sagemaker.readthedocs.io/en/v2/api/training/estimators.html#sagemaker.estimator.Estimator for all options
         fit_kwargs:
             Any extra arguments needed to pass to fit.
-            Please refer to https://sagemaker.readthedocs.io/en/stable/api/training/estimators.html#sagemaker.estimator.Framework.fit for all options
+            Please refer to https://sagemaker.readthedocs.io/en/v2/api/training/estimators.html#sagemaker.estimator.Estimator.fit for all options
         extra_ag_args: Optional[Dict[str, Any]], default = None
-            Additional entries to merge into ``ag_args.pkl``. Use this to ship caller-specific metadata to the
-            train script (e.g. ``predict_after_fit``, or ``id_column`` / ``timestamp_column`` for time series).
+            Additional entries to merge into ``ag_args.json``. Use this to ship caller-specific metadata to the
+            train script (e.g. ``predict_after_fit``, ``save_predictor``, or ``id_column`` /
+            ``timestamp_column`` for time series).
         """
         if data_channels.get("train_data") is None:
             raise ValueError("`data_channels['train_data']` is required.")
@@ -251,7 +251,7 @@ class SagemakerBackend(Backend):
             logger.log(20, f"Training with framework_version=={framework_version}")
 
         if not job_name:
-            job_name = sagemaker.utils.unique_name_from_base(CLOUD_RESOURCE_PREFIX)
+            job_name = sagemaker.utils.unique_name_from_base(self.resource_prefix)
 
         if instance_count == "auto":
             instance_count = 1
@@ -314,7 +314,7 @@ class SagemakerBackend(Backend):
                     f"`predictions_path` must be a full S3 URL ending in '.csv' or '.parquet' "
                     f"(e.g. 's3://bucket/key/predictions.parquet'), got {predictions_path!r}."
                 )
-        ag_args_path = os.path.join(self.local_output_path, "utils", "ag_args.pkl")
+        ag_args_path = os.path.join(self.local_output_path, "utils", "ag_args.json")
         self.prepare_args(path=ag_args_path, **ag_args)
         inputs = self._upload_fit_artifact(
             data_channels=data_channels,
@@ -338,7 +338,7 @@ class SagemakerBackend(Backend):
             volume_size=volume_size,
             framework_version=framework_version,
             py_version=py_version,
-            base_job_name="autogluon-cloudpredictor-train",
+            base_job_name=f"{self.resource_prefix}-train",
             output_path=output_path,
             code_location=code_location,
             inputs=inputs,
@@ -387,7 +387,7 @@ class SagemakerBackend(Backend):
             If None, will deploy the most recent trained predictor trained with `fit()`.
         endpoint_name: str
             The endpoint name to use for the deployment.
-            If None, CloudPredictor will create one with prefix `ag-cloudpredictor`
+            If None, AutoGluon Cloud creates one with a predictor- or model-specific prefix.
         framework_version: str, default = `latest`
             Inference container version of autogluon.
             If `latest`, will use the latest available container version.
@@ -400,7 +400,7 @@ class SagemakerBackend(Backend):
         custom_image_uri: Optional[str], default = None,
             Custom image to use to deploy endpoint with.
             If not specified, with use official DLC image:
-            https://github.com/aws/deep-learning-containers/blob/master/available_images.md#autogluon-inference-containers
+            https://aws.github.io/deep-learning-containers/reference/available_images/#autogluon
         volume_size: int, default = None
            The size, in GB, of the ML storage volume attached to individual inference instance associated with the production variant.
            Currenly only Amazon EBS gp2 storage volumes are supported.
@@ -409,10 +409,10 @@ class SagemakerBackend(Backend):
             To be noticed, the function won't return immediately because there are some preparations needed prior deployment.
         model_kwargs: dict, default = dict()
             Any extra arguments needed to initialize Sagemaker Model
-            Please refer to https://sagemaker.readthedocs.io/en/stable/api/inference/model.html#model for all options
+            Please refer to https://sagemaker.readthedocs.io/en/v2/api/inference/model.html#model for all options
         deploy_kwargs:
             Any extra arguments needed to pass to deploy.
-            Please refer to https://sagemaker.readthedocs.io/en/stable/api/inference/model.html#sagemaker.model.Model.deploy for all options
+            Please refer to https://sagemaker.readthedocs.io/en/v2/api/inference/model.html#sagemaker.model.Model.deploy for all options
         fm_serve_config: Optional[Dict[str, Any]], default = None
             Configuration dict passed to the FM serve script via the AG_FM_SERVE_CONFIG env var.
         inference_mode: {"realtime", "serverless"}, default = "realtime"
@@ -434,7 +434,7 @@ class SagemakerBackend(Backend):
             # Needed to infer the container image (CPU vs GPU) downstream — serverless is CPU-only.
             instance_type = "ml.m5.2xlarge"
         if not endpoint_name:
-            endpoint_name = sagemaker.utils.unique_name_from_base(CLOUD_RESOURCE_PREFIX)
+            endpoint_name = sagemaker.utils.unique_name_from_base(self.resource_prefix)
 
         # Resolve container image
         if custom_image_uri:
@@ -539,6 +539,13 @@ class SagemakerBackend(Backend):
         )
         deploy_kwargs = copy.deepcopy(deploy_kwargs or {})
         self._resolve_tags(deploy_kwargs, extra_tags)
+        inference_ami_version = infer_sagemaker_ami_version(
+            custom_image_uri,
+            instance_type,
+            image_scope="inference",
+        )
+        if inference_ami_version is not None:
+            deploy_kwargs.setdefault("inference_ami_version", inference_ami_version)
 
         instance_kwargs = {
             "instance_type": instance_type,
@@ -793,7 +800,7 @@ class SagemakerBackend(Backend):
             If `custom_image_uri` is set, this argument will be ignored.
         job_name: str, default = None
             Name of the launched training job.
-            If None, CloudPredictor will create one with prefix ag-cloudpredictor.
+            If None, AutoGluon Cloud creates one with a predictor- or model-specific prefix.
         instance_count: int, default = 1,
             Number of instances used to do batch transform.
         instance_type: str, default = 'ml.m5.2xlarge'
@@ -814,14 +821,14 @@ class SagemakerBackend(Backend):
             If `persist` is `False`, file would first be downloaded to this path and then removed.
         model_kwargs: dict, default = dict()
             Any extra arguments needed to initialize Sagemaker Model
-            Please refer to https://sagemaker.readthedocs.io/en/stable/api/inference/model.html#model for all options
+            Please refer to https://sagemaker.readthedocs.io/en/v2/api/inference/model.html#model for all options
         transformer_kwargs: dict
             Any extra arguments needed to pass to transformer.
-            Please refer to https://sagemaker.readthedocs.io/en/stable/api/inference/transformer.html#sagemaker.transformer.Transformer for all options.
+            Please refer to https://sagemaker.readthedocs.io/en/v2/api/inference/transformer.html#sagemaker.transformer.Transformer for all options.
         transform_kwargs:
             Any extra arguments needed to pass to transform.
             Please refer to
-            https://sagemaker.readthedocs.io/en/stable/api/inference/transformer.html#sagemaker.transformer.Transformer.transform for all options.
+            https://sagemaker.readthedocs.io/en/v2/api/inference/transformer.html#sagemaker.transformer.Transformer.transform for all options.
 
         Returns
         -------
@@ -898,7 +905,7 @@ class SagemakerBackend(Backend):
             If `custom_image_uri` is set, this argument will be ignored.
         job_name: str, default = None
             Name of the launched training job.
-            If None, CloudPredictor will create one with prefix ag-cloudpredictor.
+            If None, AutoGluon Cloud creates one with a predictor- or model-specific prefix.
         instance_count: int, default = 1,
             Number of instances used to do batch transform.
         instance_type: str, default = 'ml.m5.2xlarge'
@@ -919,14 +926,14 @@ class SagemakerBackend(Backend):
             If `persist` is `False`, file would first be downloaded to this path and then removed.
         model_kwargs: dict, default = dict()
             Any extra arguments needed to initialize Sagemaker Model
-            Please refer to https://sagemaker.readthedocs.io/en/stable/api/inference/model.html#model for all options
+            Please refer to https://sagemaker.readthedocs.io/en/v2/api/inference/model.html#model for all options
         transformer_kwargs: dict
             Any extra arguments needed to pass to transformer.
-            Please refer to https://sagemaker.readthedocs.io/en/stable/api/inference/transformer.html#sagemaker.transformer.Transformer for all options.
+            Please refer to https://sagemaker.readthedocs.io/en/v2/api/inference/transformer.html#sagemaker.transformer.Transformer for all options.
         transform_kwargs:
             Any extra arguments needed to pass to transform.
             Please refer to
-            https://sagemaker.readthedocs.io/en/stable/api/inference/transformer.html#sagemaker.transformer.Transformer.transform for all options.
+            https://sagemaker.readthedocs.io/en/v2/api/inference/transformer.html#sagemaker.transformer.Transformer.transform for all options.
 
 
         Returns
@@ -1015,9 +1022,9 @@ class SagemakerBackend(Backend):
             return load_pd.load(local_path)
 
     def _download_ag_args_from_job(self) -> Dict[str, Any]:
-        """Fetch and unpickle the ``ag_args.pkl`` that was uploaded as the ``ag_args`` channel.
+        """Fetch and parse the ``ag_args.json`` that was uploaded as the ``ag_args`` channel.
 
-        Each training job carries the exact pickle it was launched with as an input channel,
+        Each training job carries the exact config it was launched with as an input channel,
         making this the authoritative source — independent of local-disk lifetime.
         """
         job_name = self._fit_job.job_name
@@ -1029,12 +1036,12 @@ class SagemakerBackend(Backend):
             f"Training job {job_name!r} has no `ag_args` input channel — cannot recover predictions_path."
         )
         bucket, key = s3_path_to_bucket_prefix(ag_args_uri)
-        assert key.endswith(".pkl"), f"Expected ag_args channel to point to a .pkl file, got {ag_args_uri!r}"
+        assert key.endswith(".json"), f"Expected ag_args channel to point to a .json file, got {ag_args_uri!r}"
         with tempfile.TemporaryDirectory(prefix="ag_args_") as tmpdir:
             local_path = os.path.join(tmpdir, os.path.basename(key))
             self.sagemaker_session.boto_session.client("s3").download_file(bucket, key, local_path)
-            with open(local_path, "rb") as f:
-                return pickle.load(f)
+            with open(local_path, "r") as f:
+                return json.load(f)
 
     def _construct_ag_args(self, predictor_init_args, predictor_fit_args, leaderboard, **kwargs):
         config = dict(
@@ -1251,7 +1258,7 @@ class SagemakerBackend(Backend):
         predictor_path = self._upload_predictor(predictor_path, cloud_key_prefix + "/predictor")
 
         if not job_name:
-            job_name = sagemaker.utils.unique_name_from_base(CLOUD_RESOURCE_PREFIX)
+            job_name = sagemaker.utils.unique_name_from_base(self.resource_prefix)
 
         if test_data_image_column is not None:
             logger.warning("Batch inference with image modality could be slow because of some technical details.")
@@ -1404,7 +1411,7 @@ class SagemakerBackend(Backend):
         self.sagemaker_session = setup_sagemaker_session()
         self._region = self.sagemaker_session.boto_region_name
         if hasattr(self, "_endpoint_saved") and self._endpoint_saved is not None:
-            self.endpoiont = self.attach_endpoint(self._endpoint_saved)
+            self.attach_endpoint(self._endpoint_saved)
             self._endpoint_saved = None
         self._fit_job.session = self.sagemaker_session
         for job in self._batch_transform_jobs: