AttributeError: 'Session' object has no attribute 'settings'

[vektaBastien]

Describe the bug
Sagemaker tries to access the settings attribute of the sagemaker.modules.Session()wich does not exists.

To reproduce

import sagemaker
import boto3
from sagemaker.modules import Session
from sagemaker.pytorch import PyTorch

boto_session = boto3.Session(profile_name="vekta-production")
sagemaker_session = Session(boto_session=boto_session)

estimator = PyTorch(
    entry_point="train.py",
    source_dir=".",
    role="arn:aws:iam::261877606600:role/SageMakerRoleu",
    instance_count=1,
    instance_type="ml.t3.medium",
    framework_version="1.12.0",
    py_version="py38",
    sagemaker_session=sagemaker_session,
)
estimator.fit()

Screenshots or logs
logs:

│ /home/bastien/python_env/base/lib/python3.12/site-packages/sagemaker/estimator.py:967 in         │
│ _prepare_for_training                                                                            │
│                                                                                                  │
│    964 │   │   │   │   script = self.uploaded_code.script_name                                   │
│    965 │   │   │   │   self.code_uri = self.uploaded_code.s3_prefix                              │
│    966 │   │   │   else:                                                                         │
│ ❱  967 │   │   │   │   self.uploaded_code = self._stage_user_code_in_s3()                        │
│    968 │   │   │   │   code_dir = self.uploaded_code.s3_prefix                                   │
│    969 │   │   │   │   script = self.uploaded_code.script_name                                   │
│    970                                                                                           │
│                                                                                                  │
│ /home/bastien/python_env/base/lib/python3.12/site-packages/sagemaker/estimator.py:1058 in        │
│ _stage_user_code_in_s3                                                                           │
│                                                                                                  │
│   1055 │   │   │   dependencies=self.dependencies,                                               │
│   1056 │   │   │   kms_key=kms_key,                                                              │
│   1057 │   │   │   s3_resource=self.sagemaker_session.s3_resource,                               │
│ ❱ 1058 │   │   │   settings=self.sagemaker_session.settings,                                     │
│   1059 │   │   )                                                                                 │
│   1060 │                                                                                         │
│   1061 │   def _assign_s3_prefix(self, key_prefix=""):                                           │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯
AttributeError: 'Session' object has no attribute 'settings'

System information
A description of your system. Please provide:

• SageMaker Python SDK version: 2.245.0
• Framework name (eg. PyTorch) or algorithm (eg. KMeans): Pytorch
• Framework version: 1.12.0-cpu
• Python version:3.12.3
• CPU or GPU:CPU
• Custom Docker image (Y/N):N

Additional context
None

[Aditi2424]

The issue you're encountering is due to incompatible usage between the SageMaker SDK Estimator APIs and the new sagemaker.modules Session object introduced in recent versions of the SageMaker Python SDK.

Fix:
Update your import to use the classic Session

from sagemaker.session import Session