Cloud computing often plays a crucial role in developing and deploying intensive artificial intelligent applications by providing the necessary infrastructure and scalability. Amazon Web Services (AWS*) is one of the most comprehensive and broadly adopted Cloud Service Providers (CSP), with hundreds of services to build out any application on the cloud. AWS offers a chance for developers and businesses to innovate at a rapid pace with lower costs and greater agility by utilizing the many services offered within the platform, such as purpose-built databases and tools for a variety of applications, including machine learning and artificial intelligence.
By implementing pre-built optimizations and tools for a variety of applications and use cases on AWS, developers can further increase model efficiency and innovate at an even greater pace on popular hardware technologies. Finding and implementing the right tools and optimizations that benefit your project can be time-consuming and resource-intensive. Simplifying the implementation of these optimizations with extensive documentation and guides can help developers overcome the pain of adding new architectures to their code.
What are Intel® Cloud Optimization Modules?
Built with production AI developers in mind, Intel® Cloud Optimization Modules offer a suite of cloud-native open-source reference architectures to further maximize the potential of cloud-based solutions that integrate seamlessly with AI workloads. With these modules, developers can implement Intel-optimized AI solutions to increase the efficiency of their workloads for optimal performance on Intel CPU and GPU technologies.
The cloud optimization modules are available for popular cloud platforms like AWS, with specifically built tools to complement and enhance the cloud experience on AWS with relevant codified Intel AI software optimizations. These optimizations offer many key benefits in powering AI solutions with end-to-end AI software and optimizations for a variety of use cases, from computer vision to natural language processing and more.
Each module features a content package that includes the open-source GitHub* repository with all the documentation and a whitepaper with more information on the module and what it entails. A cheat sheet is also available in the content packages highlighting the most relevant code for each module, video series, hands-on walkthroughs on how to implement the architectures, and an option to attend office hours for any specific implementation questions.
Intel Cloud Optimization Modules for AWS
There are several Intel Cloud Optimization Modules available for AWS, including optimizations for SageMaker*, Amazon Elastic Kubernetes*, and other popular AWS tools. You can learn more about these optimization modules available for AWS below:
GPT2-Small Distributed Training
Generative pre-trained transformer (GPT) models are popular GenAI applications in a variety of spaces. Building small, large language models (LLM) is often sufficient in many use cases as the smaller models are quicker to build and easier to deploy. This module teaches developers how to fine-tune a GPT2-small (124M parameter) model on a cluster of Intel® Xeon® CPUs on AWS for high-performance distributed training. The module goes through the entire lifecycle of fine-tuning an LLM on a configured AWS cluster while capitalizing on software optimizations and frameworks like the Intel® Extension for PyTorch* and oneAPI Collective Communications Library (oneCCL) to accelerate the fine-tuning process and boosting model performance in an efficient multi-node training environment. The result is an LLM on AWS that can generate words trained on your specific task and dataset for your use case.
XGBoost on SageMaker
Amazon SageMaker is a popular tool for developing, training and deploying machine learning applications on AWS with integrated Jupyter notebook instances and common optimized machine learning algorithms for more efficient model development. By working through this module, you can learn how to inject your custom training and inference code into a prebuilt SageMaker pipeline while enabling the Intel® AI Tools for accelerated models. This module utilizes the Intel® Optimization for XGBoost* to accelerate a complete and custom end-to-end machine learning pipeline on SageMaker. The XGBoost oneDAL container includes the oneAPI Data Analytics Library to speed up model algorithms, while the Lambda container contains all the necessary components to build custom AWS Lambda functions with XGBoost and Intel oneDAL optimizations.
XGBoost on Kubernetes
Amazon Elastic Kubernetes Services (EKS) makes it easy for developers to start, run, and scale Kubernetes applications on AWS with an automatically managed service. This module facilitates developers in building and deploying accelerated AI applications on AWS with EKS and Intel® AI Tools. Developers can learn how to build an accelerated Kubernetes cluster that utilizes the Intel® Optimization for XGBoost* for AI workloads with Intel oneDAL optimizations. Along with EKS, the module also utilizes the Amazon Container Registry* (ECR), Amazon Elastic Compute Cloud* (EC2), and Elastic Load Balancer (ELB).
Enhance your AI projects on AWS with Intel Cloud Optimization Modules by utilizing Intel optimizations and containers for popular tools. You can learn how to build accelerated models on your favorite AWS tools and services and implement powerful software optimizations to take your projects to the next level. Unlock the potential of your AWS projects through these modules, and sign up for office hours if you have any questions about your implementation!
We encourage you to check out Intel’s other AI Tools and Framework optimizations and learn about the unified, open, standards-based oneAPI programming model that forms the foundation of Intel’s AI Software Portfolio. Also, check out the Intel Developer Cloud to try out the latest AI hardware and optimized software to help develop and deploy your next innovative AI projects!
AI/ML Technical Software Product Marketing Manager at Intel. MBA, Engineer, and previous two-time startup founder with a passion for all things AI and new tech.