NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #1039 most downloaded on PyPI
Open source library for training and deploying models on Amazon SageMaker.
Last release 6 days ago
18 Sep 2026
Ships on a steady schedule
a new release about every 2 weeks
Nearly every release is documented
notes for 58 of the last 60 stable releases
3 versions withdrawn
withdrawn after publishing
9 years old
691 releases · first in 2017
update boto3 minor version >= 1.20.18
Hugging Face Transformers 4.12 for Pt1.9/TF2.5
One column per quarter.
CompilationStep support for Sagemaker Pipelines
deprecate Serverless Lambda model-predictor
Update estimator docstrings to add Fast File Mode
expose num_clusters parameter for clarify shap in shapconfig
HuggingFace image_uri generation for inference
Add image_uris.retrieve() support for AutoGluon
modify RLEstimator to use newly generated Ray image (1.6.0)
warn for deprecation - Lambda model-predictor
Update timeouts for integ tests from 20 to 40
HF estimator attach modified to work with py38
support configurable retry for pipeline steps
Hugging Face Transformers 4.10 for Pt1.8/TF2.4 & Transformers 4.11 for PT1.9&TF2.5
Update TRAINING_DEFAULT_TIMEOUT_MINUTES to 40 min
Add support for Hugging Face 4.10.2
fix feature store ingestion via data wrangler test
### Bug Fixes and Other Changes * update feature request label * update bug template
ParamValidationError when scheduling a Clarify model monitor
support maps in step parameters
add checks for ExecutionRole in UserSettings, adds more unit tests
Info about offline s3 bucket key when creating feature group
Nothing published for this version
unit tests for KIX and remove regional calls to boto
update experiment config doc on fit method
### Bug Fixes and Other Changes * pin docker to 5.0.0
Add KIX account for SM XGBoost 1.2-2 and 1.3-1
support displayName and description for pipeline steps
Remove stale S3DownloadMode from test_session.py
Algorithms region launch on KIX
deprecate tag logic on Association
Nothing published for this version
add pytorch 1.5.1 eia configuration
support tuning step parameter range parameterization + support retry strategy in tuner
clarify that default_bucket creates a bucket
sklearn integ tests, remove swallowing exception on feature group delete attempt
revert #2251 changes for sklearn processor
processors that support multiple Python files, requirements.txt, and dependencies.
sagemaker.serverless integration testadd LambdaStep support for SageMaker Pipelines
null checks for uploaded_code and entry_point
Rename PredictorBase.delete_endpoint as PredictorBase.delete_predictor
PredictorBase.delete_endpoint as PredictorBase.delete_predictororder of populating container list
Set flag when debugger is disabled
Adding serial inference pipeline support to RegisterModel Step
default time for compilation jobs
remove upsert from test_workflow
add support for SageMaker workflow tuning step
smddp 1.2.1 release note / convert md to rst
handle tags when upsert pipeine
revert "fix: jsonGet interpolation issue 2426 + allow step depends on pass in step instance (#2477)"
support job_name_prefix for Clarify
update to include new batch_get_record api call
Add HF transformer version 4.6.1
Add support for Callback steps in model building pipelines
support endpoint_name_prefix, seed and version for Clarify
add xgboost framework version 1.3-1
default value removed if zero for integer param
support for custom pipeline execution name
support RetryStrategy for training jobs
add xgboost framework version 1.2-2
Your coding agent can read these notes before it upgrades. Set up the MCP server →