Advertisement

Intel offers 34 AI reference kits

Intel has introduced 34 open-source AI reference kits, designed to accelerate AI development and deployment across multiarchitecture environments.

Intel Corp. has now released a total of 34 open-source AI reference kits for developers and data scientists, thanks to a years-long collaboration with Accenture. The first set of AI reference kits were introduced last year.

The reference kits are designed to accelerate artificial intelligence (AI) deployment and make AI more accessible to multiarchitecture on-premises, cloud and edge environments. Each kit includes model code, training data, instructions for the machine-learning pipeline, libraries and oneAPI components to optimize AI.

The reference kits are built on oneAPI, an open, standards-based, heterogeneous programming model and Intel’s end-to-end AI software portfolio, including the Intel AI Analytics Toolkit and the Intel Distribution of OpenVINO toolkit. The “reference kits enable AI developers to streamline the process of introducing AI into their applications, enhancing existing intelligent solutions and accelerating deployment,” which results in a shorter, more productive workflow versus a traditional model development workflow, said Intel.

Chart showing workflow time savings of Intel's AI reference kits.

Intel’s AI reference kits reduce time to solution. (Source: Intel Corp.)

The preconfigured kits can be used for a range of solutions across industries including consumer products, energy and utilities, financial services, health and life sciences, manufacturing, retail and telecommunications. A few of those include:

  • An AI reference kit, designed to set up interactions with an enterprise conversational AI chatbot, delivering up to 45% faster inferencing with oneAPI optimizations.
  • The AI reference kit for automating visual quality control inspections for life sciences demonstrated training up to 20% faster and inferencing 55% faster for visual defect detection with oneAPI optimizations.
  • An AI reference kit for predictive asset health analytics provides up to a 25% increase in prediction accuracy.

Intel said these kits can reduce the time to solution from weeks to days, by enabling developers and data scientists to train models faster and at lower cost, thanks to the open-source environment.

Select kits will continue to be updated, Intel said. These kits include visual quality inspection, enterprise conversational AI chatbot setup, predictive asset health analytics, medical imaging diagnostics, document automation and AI-structured data generation. They can be downloaded for free at the Intel web page or on GitHub.

Advertisement



Learn more about Intel Corp.

Leave a Reply