Pipelines and pipeline runs are introduced.
- Services: Data Science
- Release Date: Jan. 25, 2023
Machine learning pipelines are a crucial component of the modern data science workflow. They help automate the process of building, …
73 Release Notes | Page 1 of 2
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.
Added --job-info
option to …
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 in Cloud …
The following changes were made in ADS 2.7.0:
Fixed a bug in GenericModel.prepare
. The .model-ignore
file wasn't included in …
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.
ads.dataset.helper
to support Python …Accelerated Data Science added support for large models that are models with artifacts between 2 and 6 GB. An Object Storage …
See the …
The following changes were made in this version.
prepare_save_deploy()
method to the GenericModel
class. Now you can prepare model …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 in Cloud …
You can now use flexible compute shapes for model deployments.
For APIs, see CreateModelDeployment, and ModelDeploymentInstanceShapeConfigDetails.
For more …
The following changes were made in this version.
Added from_model_deployment()
method to the GenericModel
class. Now, you can load a …
The Environment Explorer tab list view has been updated with:
You can now build and use your own container for use when you create a job and job runs.
For …
The following changes were made in this version.
BDSSecretKeeper
to store and save configuration parameters to connect to Big …The following conda environments are introduced:
Sklearn
models.NotebookRuntime
when using non-ASCII encoding. …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 …
Job
creation.ADSDataset get_recommendations
causing an HTML is not defined
exception.The following conda environments are introduced:
You can now monitor the health, capacity, and performance of Data Science job runs by using metrics, alarms, and notifications.
For …
storage_options
parameter in ADSDataset
.to_hdf()
.overwrite_script
or overwrite_archive …
These enhancements were made to the JupyterLab environment:
The JuypterLab Git extension is now available in notebook sessions. This change came with the latest Notebook Virtual Machine (NBVM) so you …
The following conda environments are introduced:
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.
For more …
Fixed a bug in model artifact prepare()
, reload()
, and prepare_generic_model()
resulting in an ``ONNXRuntimeError`` caused by the mismatched …
Data Labeling
ADS
DaskSeries
class was deprecated.These features were added in ADS v2.5.3:
ADS
Moved fastavro
, pandavro,
and openpyxl
to an optional dependency.
Data Labeling …
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. Using …
Accelerated Data Science v2.5.2 is released including improved model introspection functionality and the ability to manage credentials with the Vault service.
Additional …
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 each …
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 allow data scientists …
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 CPUs and GPUs. …
This new release of the model catalog is now available. It includes these enhancements:
Automatical extraction …