NXP Semiconductors has announced the addition of two new tools to its eIQ development software for artificial intelligence and machine learning. This expansion includes GenAI Flow with fine-tuning for retrieval-augmented generation (RAG) and the eIQ Time Series Studio, aiming to facilitate the deployment of AI solutions at the edge, ranging from small microcontrollers (MCUs) to powerful application processors (MPUs).
New Tools to Simplify Edge AI Development
The eIQ Time Series Studio introduces an automated machine learning workflow that simplifies the development and deployment of time series-based models on MCU-class devices, such as the MCX series of microcontrollers and the i.MX RT series of crossover MCUs. This tool enables the creation of AI models tailored to analyze sequential data, such as voltage, temperature, or vibration signals, accelerating the development of applications like anomaly detection and real-time data classification.
On the other hand, GenAI Flow facilitates the implementation of generative AI models on edge devices. Designed for NXP’s i.MX family of MPUs, it allows the training of large language models (LLMs) on domain-specific contextual data, ideal for applications requiring natural language interaction. For example, a household appliance equipped with an LLM trained on its user manual could answer questions from the user about the device’s use and maintenance.
Edge AI: A Practical and Efficient Solution
Implementing AI at the edge offers numerous advantages, including reduced latency, improved privacy, and lower energy consumption. By expanding its eIQ Toolkit, NXP enables developers to access a wide range of models, from generative AI to time series and vision models, compatible with an extensive array of edge processors.
“AI is key to a world that anticipates and automates user needs, but its development must be viable for deployment at the edge. With ready-to-use tools suitable for both small AI models on MCUs like the MCX portfolio and generative models on more powerful devices like the i.MX 95 application processor, NXP offers unprecedented options for developers across a wide range of markets,” said Charles Dachs, senior vice president and general manager of NXP Semiconductors.
Advanced Technology for AI Developers Without Data Science Experience
The eIQ Time Series Studio significantly reduces development time by providing an intuitive interface that enables the creation of optimized models for anomaly detection, classification, and regression without requiring advanced knowledge in data science. The platform allows developers to visualize, analyze, and automatically generate models tailored to performance, memory, and accuracy criteria, as well as integrate the ability to merge data from multiple sensors.
Furthermore, GenAI Flow incorporates a retrieval-augmented generation (RAG) methodology, which safely fine-tunes models on domain-specific data and private information without compromising security. This functionality allows users to customize LLMs according to the desired task, optimizing them for secure and efficient deployment at the edge.
With these new tools, NXP reinforces its commitment to making edge AI accessible and practical for developers across various industries. Interested parties can learn more and access the latest version of the eIQ development environment on NXP’s website, where they will also find a technical document on implementing generative AI at the edge.
via: NXP