sagemaker-python-sdk

Features

• TGI 1.4.5
• optimum 0.0.21
• Add TF 2.14 Graviton Inference support
• JumpStart alternative config parsing

Bug Fixes and Other Changes

• Add back serialization for automatic speech recognition
• bump apache-airflow version to 2.8.4
• Add supported task types to schema builder omission
• remove trailing slash when uploading to S3 with dataset_builder.to_csv_file
• Update Collaborator Check workflow to check for users which are part of collaborator team
• forward network_isolation parameter to Estimators when False
• Flaky slow test
• Revert "Test SM PySDK Variations"

Features

• add support to clarify.py for time series explainability jobs

Bug Fixes and Other Changes

• remove pytorch test for deprecated function
• skip test_experiment_analytics to unblock release
• Create workflow module scoped sagemaker_session to resolve test race condition
• Simplify how we process test dependencies, which are supposed to include all extras.
• skip failing feature store search integ test
• skip failing pt test
• list jumpstart models with invalid version strings
• urge customers to install latest version

Features

• Update SM Python SDK for PT 2.2.0 SM DLC

Bug Fixes and Other Changes

• Create custom tarfile extractall util to fix backward compatibility issue
• Upgrade smp to version 2.2
• Enhance model builder selection logic to include model size

Features

• pin dll version to support python3.11 to the sdk
• instance specific jumpstart host requirements
• Add TensorFlow 2.14 image configs
• Add AutoMLV2 support
• Support selective pipeline execution between function step and regular step
• Add new Triton DLC URIs

Bug Fixes and Other Changes

• Skip No Canvas regions for test_deploy_best_candidate
• make sure gpus are found in local_gpu run
• Bump Apache Airflow version to 2.8.2
• properly close sagemaker config file after loading config
• remove enable_network_isolation from the python doc

Documentation Changes

• Add doc for new feature processor APIs and classes

Features

• Prepend SageMaker Studio App Type to boto3 User Agent string
• TGI optimum 0.0.18 (general+llm)
• TGI 1.4.2

Bug Fixes and Other Changes

• tolerate vulnerable old model for integ test and temporarily skip test_list_jumpstart_models_script_filter
• add missing regions to pytorch config
• Add validation for sagemaker version on remote job
• fixed implementation of fail_on_violation for transform with monitoring

Features

• ModelBuilder to fetch local schema when no SchemaBuilder present.
• AutoGluon 1.0.0 image_uris update

Bug Fixes and Other Changes

• skip pytorch training compiler integ test
• add fixes for tarfile extractall functionality PEP-721
• Fix telemetry image uri option logic for ModelBuilder
• Add telemetry metrics on usage of default images for ModelBuilder
• Fix error message typo
• Add "distribution" parameter into record_set

Bug Fixes and Other Changes

• Add PT 2.1 as a supported framework for the smdistributed distribution
• Enable private docker registry support for ModelBuilder
• HF PT 2.1 Image Configs

Documentation Changes

• add setup commands for documentation generation

Features

• Introduce HF Transformers to ModelBuilder
• retrieve jumpstart estimator and predictor without specifying model id (infer from tags)

Bug Fixes and Other Changes

• SMP PT upgrade to 2.1
• Fetch HF metadata only when explicit type is not selected
• relax upper bound for urllib dependency

Bug Fixes and Other Changes

• update get_execution_role_arn from metadata file if present

Bug Fixes and Other Changes

• update get_execution_role_arn from metadata file if present

Features

• Logic to detect hardware GPU count and aggregate GPU memory size in MiB

Bug Fixes and Other Changes

• fixed create monitoring schedule failing after validation error
• Support PipelineVariable for ModelQualityCheckConfig attributes
• TGI NeuronX 0.0.17

Features

• Support selective pipeline execution for function step

Bug Fixes and Other Changes

• remove fastapi and uvicorn dependencies
• Support using PipelineDefinitionConfig in local mode
• update get_execution_role_arn from metadata file if present
• update image_uri_configs 01-24-2024 06:17:33 PST
• Add validation for empty ParameterString value in start local pipeline

Bug Fixes and Other Changes

• Merge branch 'master-rba' into local_merge
• Fix user agent tag issue
• update image_uri_configs 12-13-2023 14:04:54 PST
• update image_uri_configs 12-13-2023 12:23:06 PST

Deprecations and Removals

• remove explicit partitions key requirement on pysdk side.

Features

• add non-repeating config logger
• Add SageMaker Geospatial verison 1.x images
• TGI 1.2.0 Image Uri
• support model.register() with triton model
• Enable retrieving function step results for local mode

Bug Fixes and Other Changes

• TGI 1.3.1
• excessive jumpstart instance type logging
• Support local mode for remote function
• Session.download_data can not download nested objects
• Fix writing into non-closed file with git clone command
• mitigation of xgboost container incompatibility with new version
• update image and hardware validation with inf and graviton
• remove two setuptools deprecations
• minor jumpstart dev ex improvements
• save utils fix
• Correct DJL neuronx regions
• delete unused file inference-experience-dev-tester.sh
• Fix Experiment Run integ test w.r.t unexpected boto3 version
• Bump test dependencies versions
• fast follow on js uncompressed support - ModelBuilder
• Modify Region List for Neuron Images (HF neuron/neuronx, PT Neuron)

Documentation Changes

• Mention for custom Docker Image

Features

• re:Invent 2023 keynote1

Breaking Changes

• Move StreamDeserializer to sagemaker.deserializers
• Move StringDeserializer to sagemaker.deserializers
• rename record_deserializer to RecordDeserializer
• remove "train_" where redundant in parameter/variable names
• Add BytesDeserializer
• rename image to image_uri
• rename image_name to image_uri
• create new inference resources during model.deploy() and model.transformer()
• rename session parameter to sagemaker_session in S3 utility classes
• rename distributions to distribution in TF/MXNet estimators
• deprecate update_endpoint arg in deploy()
• create new inference resources during estimator.deploy() or estimator.transformer()
• deprecate delete_endpoint() for estimators and HyperparameterTuner
• refactor Predictor attribute endpoint to endpoint_name
• make instance_type optional for Airflow model configs
• refactor name of RealTimePredictor to Predictor
• remove check for Python 2 string in sagemaker.predictor._is_sequence_like()
• deprecate sagemaker.utils.to_str()
• drop Python 2 support

Features

• add BaseSerializer and BaseDeserializer
• add Predictor.update_endpoint()

Bug Fixes and Other Changes

• handle "train_*" renames in v2 migration tool
• handle image_uri rename for Session methods in v2 migration tool
• Update BytesDeserializer accept header
• handle image_uri rename for estimators and models in v2 migration tool
• handle image_uri rename in Airflow model config functions in v2 migration tool
• update migration tool for S3 utility functions
• set _current_job_name and base_tuning_job_name in HyperparameterTuner.attach()
• infer base name from job name in estimator.attach()
• ensure generated names are < 63 characters when deploying compiled models
• add TF migration documentation to error message

Documentation Changes

• update documentation with v2.0.0.rc1 changes
• remove 'train_*' prefix from estimator parameters
• update documentation for image_name/image --> image_uri

Testing and Release Infrastructure

• refactor matching logic in v2 migration tool
• add cli modifier for RealTimePredictor and derived classes
• change coverage settings to reduce intermittent errors
• clean up pickle.load logic in integ tests
• use fixture for Python version in framework integ tests
• remove assumption of Python 2 unit test runs

Breaking Changes

• remove estimator parameters for TF legacy mode
• remove legacy TensorFlowModel and TensorFlowPredictor classes
• force image URI to be passed for legacy TF images
• rename sagemaker.tensorflow.serving to sagemaker.tensorflow.model
• require framework_version and py_version for framework estimator and model classes
• change Model parameter order to make model_data optional

Bug Fixes and Other Changes

• add v2 migration tool

Documentation Changes

• update TF documentation to reflect breaking changes and how to upgrade
• start v2 usage and migration documentation

Testing and Release Infrastructure

• remove scipy from dependencies
• remove TF from optional dependencies

Bug fixes and other changes

• update py2 warning message
• add logic to use asimov image for TF 1.14 py2

Documentation changes

• changed EFS directory path instructions in documentation and Docstrings

Features

• support training inputs from EFS and FSx

Bug fixes and other changes

• Add support for Managed Spot Training and Checkpoint support
• Integration Tests now dynamically checks AZs