Data Science

Data Science Model Deployment and Model Catalog services now support large model deployments.

Large model artifacts can be stored ...

If you have models you want to use instead of those curated by Data Science, you can bring them into ...

Deeplinking into a Notebook Session enables customers of OCI Data Science to open a notebook session at a file path ...

Data Science ML Monitoring lets you:

• Read data from Object Storage using the built-in data readers.
• Extend ...

AI quick actions makes it easy for you to browse a curated list of foundation models, and deploy, fine-tune, and ...

Key Features

• Burstable Instances for Machine Learning: Lets deployment of machine learning models on virtual machines with ...

Some Key benefits of autoscaling for model deployment include:

• Dynamic Resource Adjustment: Autoscaling automatically increases or decreases ...

The following conda environments are introduced:

• AI Anomaly Detection Operator for CPU on Python 3.8 (version 1.0)
...

Data Science notebooks now use JupyterLab version 3.6.6.

• You can now manage your containers more efficiently with our new ML Jobs APIs and Console.
• Upload and ...

The following changes were made in ADS 2.10.1:

• Releasing v1 of the Anomaly Detection Operator! The Anomaly ...

The following conda environments are introduced:

• PySpark 3.2 and Feature Store (version 3.0)
• PyTorch 2.1 for GPU ...

The following changes were made in ADS 2.10.0:

• Improved the progress bar to use the percentage completed of ...

The following changes were made in ADS 2.9.1:

• Added support for deploying LangChain application as OCI Model Deployment. ...

The following changes were made in ADS 2.9.0:

• Introducing AI Forecast Operator. Learn more about Operators in the ...

The following changes were made in ADS 2.8.11:

• Added support to mount file systems in Data Science notebook ...

You can now specify File Storage service mount points or Object Storage service buckets in notebook sessions and jobs. This ...

You can now use the notebook session lifecycle scripts to run a custom script at different notebook session lifecycle states ...

You can now configure a private endpoint in your tenancy. Use a private endpoint to access one or more notebook ...

The following changes were made in ADS 2.8.10:

• Improved the LargeArtifactUploader class to understand OCI paths to upload ...

The following changes were made in ADS 2.8.9:

• Upgraded the scikit-learn dependency to >=1.0.
• Upgraded the pandas ...

The following changes were made in ADS 2.8.8:

• Added PyTorchDistributed runtime option for Data Science jobs to add ...

The following changes were made in ADS 2.8.7:

• Removed incorrect help information of opctl commands.
• Add pre-commit ...

The following changes were made in ADS 2.8.6:

• Resolved an issue in ads opctl build-image job-local when the ...

The following changes were made in ADS 2.8.5:

• Added support for key_content attribute in ads.set_auth() for the API ...

The following changes were made in ADS 2.8.4:

• Added support for creating ADSDataset from pandas dataframe.
• Added ...

The following changes were made in ADS 2.8.3:

• Added support for custom containers (Bring Your Own Container ...

The following changes were made in ADS 2.8.2:

• Removed support for Python 3.7.

• Improved ...

• The Launcher has been updated with an icon caching mechanism and a Getting Started notebook as a separate button ...

The following changes were made in ADS 2.8.1:

• Fixed a bug for ads opctl run when --auth ...

The following changes were made in ADS 2.8.0:

• Added support for the machine learning pipelines feature.

  ...

Machine learning pipelines are a crucial component of the modern data science workflow. They help automate the process of building, ...

The following changes were made in ADS 2.7.3:

• Added support for the model version set feature.

  ...

Model versioning enables you to keep records of the different models that you've trained, and your various attempts at improving ...

We have simplified the process so that there is only one option to extend a notebook session to the maximum ...

Data Science is now available in the US Midwest (Chicago) region.

For more information about Data Science and features ...

The following changes were made in ADS 2.7.0:

• Fixed a bug in GenericModel.prepare. The .model-ignore file ...

You can connect to Data Flow and run an Apache Spark application from a Data Science notebook session. These sessions ...

The following changes were made in ADS 2.6.8 and ADS 2.6.7.

2.6.8

• Fixed a bug in ads.dataset.helper ...

Accelerated Data Science added support for large models that are models with artifacts between 2 and 6 GB. An Object ...

• Data Science notebook sessions now support new flexible compute shapes.
• Data Science jobs now support flexible compute shapes. ...

The following changes were made in this version.

• Added prepare_save_deploy() method to the GenericModel class. Now you can ...

You can now set up your notebook sessions with your often used custom environment variables and Git repos to be ...

Data Science is now available in the Mexico Central (Queretaro) region.

For more information about Data Science and features ...

You can now use flexible compute shapes for model deployments.

For APIs, see CreateModelDeployment, and ModelDeploymentInstanceShapeConfigDetails. For more ...

Example Notebooks

• Release of the new Notebook Explorer extension for JupyterLabs. This extension allows you to search for ...

The following changes were made in this version.

• Added from_model_deployment() method to the GenericModel class. Now, you ...

The Environment Explorer tab list view has been updated with:

• The Actions menu that allows you to easily ...

You can now build and use your own container for use when you create a job and job runs.

The following changes were made in this version.

• Added BDSSecretKeeper to store and save configuration parameters to connect ...

The following conda environments are introduced:

• PySpark 3.0 and Data Flow (version 4.0)
• PySpark and Data Flow ...

• Added the following framework-specific classes for fast and easy model deployment:
  • AutoMLModel
  • SKlearnModel
  • XGBoostModel
  • LightGBMModel ...

• Fixed bug in automatic extraction of taxonomy metadata for Sklearn models.
• Fixed bug in jobs NotebookRuntime when using ...

You can now choose to let the Data Science service configure your networking resources with public internet access in notebook ...

The Financial Services conda environment is an ecosystem to do portfolio optimization, stock analysis, technical analysis, and other financial econometric ...

With the PySpark 3.0 and Data Flow CPU on Python 3.7 (version 3.0) conda environment you can apply the power ...

You can use the new fast launch option to automatically select a Compute shape from predefined pool of shapes when ...

• Fixed bug in DataFlow Job creation.
• Fixed bug in ADSDataset get_recommendations causing an HTML is not defined exception. ...

The following conda environments are introduced:

• The General Machine Learning for CPU and GPU on Python 3.7 conda ...

You can now monitor the health, capacity, and performance of Data Science job runs by using metrics, alarms, and notifications. ...

• Added support for the storage_options parameter in ADSDataset .to_hdf().
• Fixed an error message to specify overwrite_script or ...

These enhancements were made to the JupyterLab environment:

• The Environment Explorer tab list view has been updated for ...

The JuypterLab Git extension is now available in notebook sessions. This change came with the latest Notebook Virtual Machine (NBVM) ...

The following conda environments are introduced:

• The Data Exploration and Manipulation for CPU v3 conda environment contains libraries ...

The Computer Vision for CPU and GPU on Python 3.7 (version 1.0) conda environment include some of the most powerful ...

You can now monitor the health, capacity, and performance of Data Science model deployments by using metrics, alarms, and notifications. ...

Fixed a bug in model artifact prepare(), reload(), and prepare_generic_model() resulting in an ``ONNXRuntimeError`` caused by the mismatched ...

Data Labeling

• Added support to read exported dataset from the consolidated export file.

ADS

...

These features were added in ADS v2.5.3:

ADS

• Moved fastavro, pandavro, and openpyxl to an ...

You can now extend your notebook session time by up to 23 hours to avoid being logged out after an ...

The Neurophysiology 1.0 for CPU healthcare focused conda environment provides the best-in-class tooling for analyzing, visualizing and exploring neurophysiological data. ...

Accelerated Data Science v2.5.2 is released including improved model introspection functionality and the ability to manage credentials with the Vault ...

The TensorFlow conda environment is an ecosystem of tools and libraries to create state-of-the-art machine learning models. You can use ...

We are introducting a new Launcher tab in JupyterLab notebook sessions. When you open the Launcher tab, you'll see that ...

Speed up your scikit-learn algorithms with the Intel Extension for Scikit-learn conda environment. It uses the Intel oneAPI Data Analytics ...

You can now access Object Storage in Pandas. The Data Science service has created the ocifs library that is based ...

The Environment Explorer has been extended by adding a new list view. This is an alternative view to the card ...

The Data Science jobs feature is introduced in ADS v2.4.0 and includes the following:

• Data Science jobs ...

The jobs feature enables you to define and run a repeatable machine learning task on a fully managed infrastructure. Jobs ...

Use​​​​​ the new PyTorch 1.9 condas for applications in computer vision and natural language processing. These conda environments are for ...

ADS v2.3.1

Model Catalog

This new release of the model catalog is now available. It includes these enhancements: ...

Data Science is now available in the Brazil Southeast (Vinhedo) region.

For more information about Data Science and features ...

This new release of the model catalog is now available. It includes these enhancements:

• New model taxonomy metadata ...

The latest version of Jupyter Lab is used when you create a new notebook. You can't choose the older version ...

Use PyPGX V21.3.1 to extract hidden insights using graph machine learning, optimized analytics algorithms, and a graph query language. PyPGX ...

The Data Science service provides seamless integration with the Data Flow service. You can develop your Spark applications in PySpark ...

Use​​​​​ the PySpark V3.0 conda to create Data Flow jobs or run PySpark locally. The PySpark version is updated from ...

The VM.Standard.E3.Flex Compute shape is now generally available in the Data Science service in certain regions. For more information, see ...

Data Science is now available in US government regions:

• US Gov West (Phoenix)

• US ...

The Data Science service provides conda environments that are specialized for different workflows. In this release, there are six new ...

Conda environments provide flexibility and reproducibility within the Data Science notebook. The Environment Explorer is GUI tool that allows you ...

Data Science is now available in the Chile (Santiago) region. For more information, see Data Science and Data Science API ...

Data Science is now available in the UK West (Newport) region. For more information, see Data Science and Data Science ...

Conda Environments

• Release of the new conda environments feature in notebook sessions. A new Environment Explorer extension is ...

You can now monitor the health, capacity, and performance of Data Science notebook sessions by using metrics, alarms, and notifications, ...

GPUs VMs (P100 and V100) are now generally available within the Oracle Cloud Infrastructure Data Science service in certain regions. ...

Data Science is now available in India South (Hyderabad), South Korea North (Chuncheon), and US West (San Jose) regions. For ...

Support for Resource Principals in Notebook Sessions

Oracle Cloud Infrastructure Data Science enables you to authenticate using your notebook ...

Notebook Session Environment

The notebook session environment includes these improvements:

• New Launcher buttons for both the notebook ...

The Notebook Session environment includes the following changes:

• The home folder is now backed by block volume. You ...

Data Science is now available in Australia Southeast (Melbourne), Canada Southeast (Montreal), Japan Central (Osaka), Netherlands Northwest (Amsterdam), and Saudi ...

The JupyterLab notebook session interface is enhanced so that the JupyterLab environment now supports:

• The Variable Inspector extension. ...

The Data Science service enables data science teams to build, train, and manage machine learning and artificial intelligence models on ...