SageMaker Python SDK 1.7.2

• bug-fix: Prediction output for the TF_JSON_SERIALIZER
• enhancement: Add better training job status report

SageMaker Python SDK 1.7.1

• bug-fix: get_execution_role no longer fails if user can't call get_role
• bug-fix: Session: use existing model instead of failing during create_model()
• enhancement: Estimator: allow for different role from the Estimator's when creating a Model or Transformer

SageMaker Python SDK 1.7.0

• feature: Transformer: add support for batch transform jobs
• feature: Documentation: add instructions for using Pipe Mode with TensorFlow

SageMaker Python SDK 1.6.1

• feature: Added multiclass classification support for linear learner algorithm.

SageMaker Python SDK 1.6.0

• feature: Add Chainer 4.1.0 support
• bug-fix: Use chainer_full_version fixture in Chainer integration tests

SageMaker Python SDK 1.5.4

• feature: Added Docker Registry for all 1p algorithms in amazon_estimator.py
• feature: Added get_image_uri method for 1p algorithms in amazon_estimator.py
• Support SageMaker algorithms in FRA and SYD regions

SageMaker Python SDK 1.5.3

• bug-fix: Can create TrainingJobAnalytics object without specifying metric_names.
• bug-fix: Session: include role path in get_execution_role() result
• bug-fix: Local Mode: fix RuntimeError handling

SageMaker Python SDK 1.5.2

• Support SageMaker algorithms in ICN region.

SageMaker Python SDK 1.5.1

• enhancement: Let Framework models reuse code uploaded by Framework estimators
• enhancement: Unify generation of model uploaded code location
• feature: Change minimum required scipy from 1.0.0 to 0.19.0
• feature: Allow all Framework Estimators to use a custom docker image.
• feature: Option to add Tags on SageMaker Endpoints

SageMaker Python SDK 1.5.0

• feature: Add Support for PyTorch Framework
• feature: Estimators: add support for TensorFlow 1.7.0
• feature: Estimators: add support for TensorFlow 1.8.0
• feature: Allow Local Serving of Models in S3
• enhancement: Allow option for HyperparameterTuner to not include estimator metadata in job
• bug-fix: Estimators: Join tensorboard thread after fitting