v1.56.3

Bug Fixes and Other Changes

• update xgboost latest image version

v1.56.2

Bug Fixes and Other Changes

• training_config returns MetricDefinitions
• preserve inference script in model repack.

Testing and Release Infrastructure

• support Python 3.7

v1.56.1.post1

Documentation Changes

• document model.tar.gz structure for MXNet and PyTorch
• add documentation for EstimatorBase parameters missing from docstring

v1.56.1.post0

Testing and Release Infrastructure

• add doc8 check for documentation files

v1.56.1

Bug Fixes and Other Changes

• add super() call in Local Mode DataSource subclasses
• fix xgboost image incorrect latest version warning
• allow output_path without trailing slash in Local Mode training jobs
• allow S3 folder input to contain a trailing slash in Local Mode

Documentation Changes

• Add namespace-based setup for SageMaker Operators for Kubernetes
• Add note about file URLs for Estimator methods in Local Mode

v1.56.0

Features

• add EIA support for TFS 1.15.0 and 2.0.0

Bug Fixes and Other Changes

• use format strings intead of os.path.join for Unix paths for Processing Jobs

v1.55.4

Bug Fixes and Other Changes

• use valid encryption key arg for S3 downloads
• update sagemaker pytorch containers to external link
• allow specifying model name when creating a Transformer from an Estimator
• allow specifying model name in create_model() for TensorFlow, SKLearn, and XGBoost
• allow specifying model name in create_model() for Chainer, MXNet, PyTorch, and RL

Documentation Changes

• fix wget endpoints
• add Adobe Analytics; upgrade Sphinx and docs environment
• Explain why default model_fn loads PyTorch-EI models to CPU by default
• Set theme in conf.py
• correct transform()'s wait default value to "False"

Testing and Release Infrastructure

• move unit tests for updating an endpoint to test_deploy.py
• move Neo unit tests to a new file and directly use the Model class
• move Model.deploy unit tests to separate file
• add Model unit tests for delete_model and enable_network_isolation
• skip integ tests in PR build if only unit tests are modified
• add Model unit tests for prepare_container_def and _create_sagemaker_model
• use Model class for model deployment unit tests
• split model unit tests by Model, FrameworkModel, and ModelPackage
• add Model unit tests for all transformer() params
• add TF batch transform integ test with KMS and network isolation
• use pytest fixtures in batch transform integ tests to train and upload to S3 only once
• improve unit tests for creating Transformers and transform jobs
• add PyTorch + custom model bucket batch transform integ test

v1.55.3

Bug Fixes and Other Changes

• remove .strip() from batch transform
• allow model with network isolation when creating a Transformer from an Estimator
• add enable_network_isolation to EstimatorBase

v1.55.2

Bug Fixes and Other Changes

• use .format instead of os.path.join for Processing S3 paths.

Testing and Release Infrastructure

• use m5.xlarge instances for "ap-northeast-1" region integ tests.

v1.55.1

Bug Fixes and Other Changes

• correct local mode behavior for CN regions