A breakthrough for AI, NuPIC empowers enterprises to easily deploy robust generative AI applications on commodity CPUs
Numenta Inc., the world leader in deploying large AI models on CPUs, announced version 2.0 of its flagship product, the Numenta Platform for Intelligent Computing (NuPIC). NuPIC empowers companies to deploy large language models (LLMs) on CPUs, offering an efficient, scalable, and secure solution.
With a focus on flexibility and real-world applications, NuPIC makes it easy for businesses to choose and deploy the right model for the right task. Whether customers want to run an existing model, fine-tune models with their proprietary data, or leverage a pre-optimized model for immediate deployment, NuPIC is designed to simplify the process and help businesses develop and power their generative AI applications.
“Why would anyone want to deploy GPUs anymore? With this latest version of NuPIC 2.0, we’ve solidified CPUs as the ideal architecture for generative AI deployments, empowering enterprises with unprecedented flexibility and simplicity,” said Subutai Ahmad, CEO of Numenta. “Whether running hundreds of models on a single server or tuning models to their performance needs, customers can leverage generative AI techniques such as RAG on NuPIC to build gen AI applications. Best of all, they can build and deploy models with complete privacy on CPUs, all within NuPIC with very little lift.”
Unlocking Generative AI Inference on CPUs
By leveraging the combination of NuPIC’s optimized models and CPUs’ inherent flexibility, NuPIC makes it possible to run large-scale generative AI models, such as Llama-2, on CPUs. With no batching required and the ability to deploy on-premise, NuPIC enhances the efficiency of AI applications and addresses privacy and security concerns associated with cloud-based solutions.
With NuPIC, enterprises can benefit from:
- Building and scaling generative AI applications on CPUs, with no deep learning expertise required.
- Leveraging robust GPT capabilities, from optimized GPT models to fine-tuning support to the ability to bring your own GPT model
- Choosing the ideal model for their use case, whether it requires GPT, embedding models, or a combination of both,
- Integrating NuPIC within their existing infrastructure, on-premise or via private cloud. No data ever leaves company walls.
- Optimal performance on Intel Xeon Processors, with results that can significantly surpass GPUs.
- Easily accessing Numenta’s team of experts to identify bottlenecks and guide them through the complex AI landscape.
NuPIC Customer Use Cases
NuPIC is gaining widespread industry adoption. Partners and customers include AI Singapore, Cequence, F5, Finch AI, Gallium Studios, and Intel.
F5 has partnered with Numenta to make Generative AI applications easier and more affordable to build. The solution is achieved by integrating NuPic into the F5 Distributed Cloud Platform to securely deliver LLM workloads for multicloud applications regardless of deployment complexity.
“Numenta enables F5 customers to run LLMs on their existing x86 based CPUs with GPU-level performance or better,” said Pranav Dharwadkar, Vice President, Product Management at F5. “This partnership is game-changing for our customers, as they are no longer held back waiting for GPUs and can benefit from running their LLMs on F5 App Stack for inference at the edge with their confidential data remaining private with RAG. With lower costs than GPUs and reduced complexity, F5 and Numenta lower the barriers to entry for GenAI applications for any company of any size.”
Cequence is using NuPIC to develop unique models, exclusively trained on the archives of client contracts. These models will provide unparalleled expertise in contract management and offer users a powerful tool that understands their unique needs and requirements. “We are excited to have formed a partnership with Numenta in our efforts to progress the application of AI in contract management. This collaboration will enable us to provide legal professionals and procurement teams with even smarter, more efficient, and user-friendly solutions,” said Martin Ragan, Chief Revenue Officer at Cequence.
Note to Editors
Numenta will showcase NuPIC 2.0 at Enterprise Generative AI Summit West Coast on May 21 and 22. Numenta CEO Subutai Ahmad will deliver a keynote address on May 21 from 11:30-12:00 p.m PT.
Explore NuPIC and Request a Demo
If you are interested in seeing how NuPIC can help you build and scale your generative AI applications on CPUs, visit our solutions page or request a demo here.
The post Numenta Unveils NuPIC 2.0 first appeared on AI-TechPark.
Source link
lol
