Data Science now has conda environments in notebook sessions.

Conda Environments

  • Release of the new conda environments feature in notebook sessions. A new Environment Explorer extension is now available in the notebook session along with the odsc conda CLI that lets you manage the lifecycle of conda environments (install, create, clone, publish, and delete) in your notebook sessions, see the Oracle AI and Data Science Blog. Also, we published video tutorial that shows you how to use conda envioronments in your notebook session.
  • Data Science conda environments are available to install in notebook sessions. Six environments are available for CPU shapes and three environments for GPU shapes, see understanding conda environments.

      CPUs:

  • Classic CPU Notebook Session Kernel
  • General Machine Learning for CPUs including new versions of Oracle Accelerated Data Science (ADS), AutoML, and MLX
  • Pyspark for Data Flow development work
  • Oracle Database
  • Data Exploration & Manipulation
  • ONNX

      GPUs:

  • Classic GPU Notebook Session Kernel
  • General Machine Learning for GPUs including new versions of ADS, AutoML, and MLX
  • NVIDIA RAPIDS 0.16

Notebook Examples

Notebook examples are now tied to Data Science Conda Environments. New notebook examples for all conda environments. Each environment has an individual `getting-started.ipynb` notebook.

AutoML and ADSModel Support of ONNX Format

New AutoML and ADSModel support for ONNX-based models. AutoML models saved as onnx objects can now be deployed using the Oracle Function service.

ADSTuner

New hyperparameter optimizer, ADSTuner, was added ADS, see the ADS SDK documentation. A notebook example, adstuner.ipynb, is available in the General Machine Learning for CPUs and GPUs environment.

New Model Artifact Version

Updates to the model artifact version were made, see creating models. ADS now captures more detailed provenance information in the runtime.yaml file, see the ADS Release Notes.

Region Additions

Data Science is now available in the Dedicated Region Cloud@Customer (Chiyoda) and UAE East (Dubai) regions. For more information, see Data Science and Data Science API.