In this recorded webinar, Numenta introduces the latest release of their AI platform, NuPIC 2.0, which allows users to build and deploy AI applications on CPUs, eliminating the need for GPUs. The session is led by Andy Xu, VP of Engineering, and Sophie Marinaro, Account Executive, who provide an overview, live demo, and Q&A session.
The webinar begins with an introduction to Numenta, founded in 2005 by Jeff Hawkins and Donna Dubinsky, with a mission to understand neuroscience and apply those insights to AI and machine learning. Numenta’s research has led to the development of the NuPIC platform, which leverages neuroscience-inspired algorithms to optimize AI models for CPU deployment.
Key benefits of using NuPIC include:
1. Cost Efficiency: NuPIC allows AI applications to run on CPUs, reducing the high costs associated with GPUs.
2. Data Privacy: The platform ensures 100% private and secure deployments, with no visibility into user data or models.
3. Scalability: NuPIC supports scalable AI applications on CPUs, making it easier to deploy large models without the need for GPUs.
4. Ease of Use: The platform integrates performant models and sample code, simplifying the deployment process for AI applications.
Andy Xu explains the architecture of NuPIC, highlighting the optimized inference server that runs large language models efficiently on CPUs. The platform supports various use cases, including customer support, question answering, and document retrieval. Andy demonstrates a document retrieval use case using a Jupyter Notebook, showcasing how to build embeddings, load data into a vector database, and perform queries using NuPIC’s optimized models.
Sophie Marinaro presents the performance comparison between NuPIC’s optimized models and traditional models running on GPUs, showing significant speed improvements. She emphasizes the simplicity and flexibility of deploying AI applications on NuPIC, making it accessible without the need for specialized hardware or extensive new learning.
The webinar concludes with a Q&A session addressing topics such as model support, system requirements, and future plans for expanding into other domains like computer vision. Attendees are encouraged to schedule a use case consultation to explore how NuPIC can benefit their specific AI projects.
Overall, the webinar provides a comprehensive overview of NuPIC 2.0, demonstrating its capabilities and advantages in building and deploying AI applications on CPUs.