v2.172.0

Features

• Add check for if TrialComponent is already associated with a Trial in Run
• Add features_to_explain to shap config

Bug Fixes and Other Changes

• Support protobuf4
• Remove unnecessary get caller identity call
• Missing JumpStart estimator args
• Add volume to partition djl_inference

Documentation Changes

• Correct runtime param
• fix wait_for_endpoint docstring

v2.171.0

Features

• Add PipelineDefinitionConfig to pipelines to toggle custom job …

Bug Fixes and Other Changes

• Upgrade DJL deepspeed versions
• Remove unused dependency protobuf3-to-dict
• skip intelligent volume_size allocation based on instance type if it is a pipeline parameter

v2.170.0

Features

• Enable customizing artifact output path

Bug Fixes and Other Changes

• Add il-central-1 support for all SM DLCs
• jumpstart async inference config predictor support
• Update CreateEdgePackagingJob resourceKey with type string

v2.169.0

Features

• Add support for tags in to_pipeline API for feature processor
• model registry integration to model cards to support model packages
• SDK Defaults - DebugHookConfig defaults in TrainingJob API
• Add segment config for Clarify

Bug Fixes and Other Changes

• Neuronx image retrieval missing sdk information

Documentation Changes

• Doc updates for SDK defaults - S3 Params, Env Variables, Disable Profiler, and DebugHookConfig

v2.168.0

Features

• Support uncompressed model upload
• Add optional monitoring_config_override parameter in suggest_baseline API
• SDK defaults add disable profiler to createTrainingJob

Bug Fixes and Other Changes

• Enable spark processing container in KIX
• Fix key prefix preventing jumpstart model repack

v2.167.0

Features

• add SageMaker FeatureStore feature processing

Bug Fixes and Other Changes

• Chore/reset cache if js model not found

v2.166.0

Features

• Add inf2 support to HuggingFaceModel
• adding resourcekey and tags for api in config for SDK defaults

Bug Fixes and Other Changes

• Remove deprecated option.s3url in favor of option.model_id
• Use sagemaker config keyword
• SDK Defaults Config - Handle config injection for None Sessions
• Fix HPO Grid Search comparison and name

v2.165.0

Features

• Add support for Deployment Recommendation ID in model.deploy(). No tagging support

Bug Fixes and Other Changes

• maketplace integs
• Add tagging assert to inference recommender integ tests
• breaking deviations in _create_sagemaker_model call

Documentation Changes

• Add missing quotation mark

v2.164.0

Features

• SDK Defaults - Environment Variables
• Update Transformers 4.28 - PyTorch 2.0.0 Training and Inference Image URI

Bug Fixes and Other Changes

• tag more integs as flaky for auto-retry
• Remove docker-compose from local requirements
• enable neo framework version support on ml_inf2 and ml_trn1

v2.163.0

Features

• Add huggingface-llm 0.8.2 dlc images

Bug Fixes and Other Changes

• Update to more actionable error message
• Loosen local reqs for PyYAML