ADS 2.8.9

• Upgraded the scikit-learn dependency to >=1.0.
• Upgraded the pandas dependency to >1.2.1,<2.1 to allow you to use ADS with pandas 2.0.
• Implemented multi-part upload in the ArtifactUploader to upload model artifacts to the model catalog.
• Fixed the "Attribute not found" error, when deploy() called twice in GenericModel.
• Fixed the fetch of the security token, when the relative path for the security_token_file is provided (used in session token-bases authentication).

ADS 2.8.8

• Added PyTorchDistributed runtime option for Data Science jobs to add support for training large language models with PyTorch.
• Added options to configure flexible shape in opctl.
• Refactored deploy() in GenericModel to prioritize the parameters.
• Fixed the opctl commands delete/cancel/watch/activate/deactivate commands to add missing parameter options.
• Fixed the opctl commands to call run to start an ML job when no YAML is specified.
• Deprecated the DatasetFactory class, and refactored the code.

ADS 2.8.7

• Added support for leveraging pools in the Data Flow applications.
• Added support for token-based authentication.
• Revised help information for opctl commands.

ADS 2.8.6

• Resolved an issue in ads opctl build-image job-local when the build of job-local would get stuck. Updated the Python version to 3.8 in the base environment of the job-local image.
• Fixed a bug that prevented the support of defined tags for Data Science job runs.
• Fixed a bug in the entryscript.sh of ads opctl that attempted to create a temporary folder in the /var/folders directory.
• Added support for defined tags in the Data Flow application and application run.
• Deprecated the old ModelDeploymentProperties and ModelDeployer classes, and their corresponding APIs.
• Enabled the uploading of large size model artifacts for the ModelDeployment class.
• Implemented validation for shape name and shape configuration details in Data Science jobs and Data Flow applications.
• Added the capability to create ADSDataset using the Pandas accessor.
• Provided a prebuilt watch command for monitoring Data Science jobs with ads opctl.
• Eliminated the legacy ads.dataflow package from ADS.

2.8.5

ADS

• Added support for key_content attribute in ads.set_auth() for the API KEY authentication.
• Fixed bug in ModelEvaluator when it returned incorrect ROC AUC characteristics.
• Fixed bug in ADSDataset.suggest_recommendations() API, when it returned an error if the target wasn't specified.
• Fixed bug in ADSDataset.auto_transform() API, when an incorrect sampling was suggested for imbalanced data.

2.8.4

ADS

• Added support for creating ADSDataset from pandas dataframe.
• Added support for multi-model deployment using Triton.
• Added support for model deployment local testing in ads opctl CLI.
• Added support in ads opctl CLI to generate starter YAML specification for the Data Science Job, Data Flow Application, Data Science Model Deployment and ML Pipeline services.
• Added support for invoking model prediction locally with predict(local=True).
• Added support for attaching customized score.py when preparing model.
• Added status check for model deployment delete/activate/deactivate APIs.
• Added support for training and verifying SparkPipelineModel in Dataflow.
• Added support for generating score.py for GPU model deployment.
• Added support for setting defined tags in Data Science jobs.
• Improved model deployment progress bar.
• Fixed bug when using ads opctl CLI to run jobs locally.
• Fixed bug in Dataflow magic when using archive_uri in dataflow config.

2.8.3

ADS

• Added support for custom containers (Bring Your Own Container or BYOC) and environment variables for ads.model.GenericModel.
• Added default values for configuring parameters in ads.model.ModelDeployment, such as default flex shape, ocpus, memory in gbs, bandwidth, and instance count.
• Added support for ads.jobs.NotebookRuntime to use directory as job artifact.
• Added support for ads.jobs.PythonRuntime and ads.jobs.GitPythonRuntime to use shell script as entrypoint.

2.8.2

ADS

• Remove support for Python 3.7.
• Improved the DataScienceMode.create() to support timeout argument and auto extract region from the signer and signer config.
• Support Jupyter Notebook as entrypoint when defining Data Science jobs with PythonRuntime and GitPythonRuntime.
• Support environment variable substitution in Data Science job names and output URI.
• Support JSON serialization of list/dictionary when assigning them as Data Science jobs environment variables.
• Support saving the notebook to output URI even if the job run failed when running a Data Science job using NotebookRuntime.
• Added job.build() method to Data Science job to load default values from environment.
• Added DataScienceJob.fast_launch_shapes() method to list fast launch shapes available for Data Science job.
• Added :doc:`HuggingFacePipelineModel class to support prepare, save, deploy and predict for HuggingFace pipelines.
• Updated Data Science job run YAML representation to include configurations inherited from the job.
• Fixed custom conda environment not showing in Data Science Job YAML specification.
• Fixed an issue where model saving was failing in notebook session without ipywidgets installed.
• Fixed "Unknown archive format" error in ads.jobs.PythonRuntime, when the source code folder name ends with "zip". List of supported archive files are: "zip", "tar.gz", "tar" and "tgz".

2.8.1

ADS

• Fixed a bug for ads opctl run when --auth flag is passed and image is built by ADS.
• Fixed a bug in GenericModel.save() when the work requests are not successfully populated.
• Fixed a bug in DataScienceModel.create() to when the provenance metadata is not provided.

2.8.0

ADS

• Added support for the machine learning pipelines feature.
• Fixed a bug in fetch_training_code_details(). When git commit is empty string, set it as None to avoid service error.
• Fixed a bug in fetch_training_code_details(). Use the folder of training_script_path as the artifact directory, instead of ..