Red Hat has announced the release of version 2.15 of its Red Hat OpenShift AI, a substantial update to their cloud-based AI and machine learning platform, which will become generally available in mid-November. The main highlight of this release is the introduction of a model registry, which provides a structured approach for managing AI models and their associated metadata and artifacts. This registry simplifies the process of sharing, versioning, deploying, and tracking models, currently offered in a technology preview state. This latest version of OpenShift AI aims to address a variety of challenges faced by data scientists and AI engineers, offering robust tools and features for efficient model management.
Enhanced Model Management Tools
One of the key enhancements in Red Hat OpenShift AI 2.15 is the introduction of data drift detection tools, which are critical for monitoring changes in the input data distributions of deployed machine learning (ML) models. Detecting significant deviations between live data and the datasets used for initial model training is essential for verifying the reliability and performance of the models in real-world applications. These tools enable data scientists to swiftly identify and address issues arising from data drift, thereby maintaining the integrity and accuracy of their ML models. Alongside these, bias detection tools from the TrustyAI open-source community have been seamlessly integrated into the platform. These tools provide the capability to monitor and ensure the fairness and unbiased nature of models during deployment.
Another notable feature in this update is the support for fine-tuning large language models (LLMs) using LoRA (low-rank adaptation), which significantly enhances the efficiency of the fine-tuning process. This feature not only allows organizations to scale their AI workloads more effectively but also reduces the associated costs and resource consumption. The platform’s support strategy now includes Nvidia NIM (interface microservices), aiming to accelerate the delivery of generative AI applications. These enhancements are indicative of Red Hat’s commitment to delivering cutting-edge AI solutions that cater to the evolving needs of enterprises.
Expanding Hardware and Security Capabilities
Red Hat OpenShift AI 2.15 also emphasizes expanded hardware support by incorporating AMD GPUs and providing access to AMD ROCm (Radeon Open Compute) workbench images. This addition facilitates model development using AMD’s GPU technology, offering greater flexibility and performance options for data scientists and AI engineers. Furthermore, the platform has introduced capabilities for serving generative AI models, including the vLLM serving runtime for KServe, a Kubernetes-based model inference platform, and KServe ModelCars, which utilize Open Container Initiative (OCI) repositories for storing and accessing model versions.
The latest update also enhances the platform’s security posture with the introduction of private/public route selection for endpoints in KServe. This feature enables organizations to direct models to internal endpoints when necessary, ensuring greater control and security over their model deployments. The integration of these security measures reflects Red Hat’s dedication to providing a secure and reliable environment for AI and ML model management.
Advancements in AI and Data Science Workflow
Version 2.15 of OpenShift AI addresses numerous challenges faced by professionals in the AI and machine learning fields. The platform comes equipped with robust tools and features designed for efficient model management, facilitating smoother operations and better collaboration among teams. This release emphasizes enhancing productivity and simplifying complex tasks, reinforcing Red Hat’s commitment to supporting the AI community with advanced and user-friendly solutions.
