Kubeflow 1.10.0 introduces significant updates aimed at enhancing machine learning workflows through improved flexibility, efficiency, and scalability. Key features include Trainer 2.0, a new UI for Model Registry, and the inclusion of Spark Operator as a core component. The release also focuses on Kubernetes and container security, provides enhancements to hyperparameter optimization, and integrates deeper with KServe. Additionally, the Model Registry now offers improved management capabilities and a new user interface, while the Training Operator and Katib receive updates facilitating better hyperparameter optimization for large language models. New features in the Kubeflow Pipelines, Notebooks, and Spark Operator further enhance performance and usability.

8m read timeFrom blog.kubeflow.org
Post cover image
Table of contents
Highlight featuresKubeflow Platform (Manifests & Security)PipelinesModel RegistryTraining Operator (Trainer) & KatibDashboard & NotebooksSpark OperatorKServeWhat comes next?How to get started with 1.10Join the CommunityWant to help?

Sort: