Cerence Introduces Pioneering Embedded Small Language Model, Purpose-Built for Automotive
Cerence has launched CaLLM™ Edge, an automotive-grade embedded small language model developed in collaboration with Microsoft. Built on Microsoft's Phi-3 family and fine-tuned with Cerence's automotive dataset, the model features 3.8 billion parameters and 4k context size. It enables car control commands, POI search, and conversational interaction without requiring internet connectivity.
The system operates in embedded-only or hybrid deployments, ensuring constant availability of generative AI features while maintaining data privacy by keeping information within the vehicle. This solution offers automakers improved assistant performance and cost efficiency while delivering cloud-like experiences regardless of connectivity.
Cerence ha lanciato CaLLM™ Edge, un modello di linguaggio embedded di grado automobilistico sviluppato in collaborazione con Microsoft. Basato sulla famiglia Phi-3 di Microsoft e perfezionato con il dataset automobilistico di Cerence, il modello dispone di 3,8 miliardi di parametri e una dimensione del contesto di 4k. Consente comandi di controllo dell'auto, ricerca di punti di interesse e interazione conversazionale senza necessitare di connettività a Internet.
Il sistema opera in modalità embedded-only o ibrida, garantendo la disponibilità costante delle funzionalità di intelligenza artificiale generativa mantenendo la privacy dei dati, tenendo le informazioni all'interno del veicolo. Questa soluzione offre ai produttori di automobili prestazioni migliori per gli assistenti e efficienza nei costi, fornendo esperienze simili al cloud indipendentemente dalla connettività.
Cerence ha lanzado CaLLM™ Edge, un modelo de lenguaje integrado de grado automotriz desarrollado en colaboración con Microsoft. Basado en la familia Phi-3 de Microsoft y ajustado con el conjunto de datos automotriz de Cerence, el modelo cuenta con 3.8 mil millones de parámetros y un tamaño de contexto de 4k. Permite comandos de control del automóvil, búsqueda de puntos de interés e interacción conversacional sin necesidad de conectividad a Internet.
El sistema opera en implementaciones solo embebidas o híbridas, garantizando la disponibilidad constante de las funciones de IA generativa mientras mantiene la privacidad de los datos al mantener la información dentro del vehículo. Esta solución ofrece a los fabricantes de automóviles un mejor desempeño del asistente y eficiencia en costos, proporcionando experiencias similares a la nube independientemente de la conectividad.
세렌스(Cerence)는 CaLLM™ Edge를 출시했습니다. 이는 마이크로소프트(Microsoft)와 협력하여 개발된 자동차 등급의 내장형 언어 모델입니다. 마이크로소프트의 Phi-3 계열을 기반으로 하고, 세렌스의 자동차 데이터셋으로 세밀하게 조정된 이 모델은 38억 개의 매개변수와 4k의 컨텍스트 크기를 가지고 있습니다. 이는 인터넷 연결 없이도 자동차 제어 명령, 관심 지점(POI) 검색 및 대화형 상호작용을 가능하게 합니다.
시스템은 내장형 전용 또는 하이브리드 배포로 작동하며, 차량 내에서 정보를 유지함으로써 데이터 프라이버시를 보장하면서 생성 AI 기능의 지속적인 가용성을 보장합니다. 이 솔루션은 자동차 제조업체에게 개선된 어시스턴트 성능과 비용 효율성을 제공하며, 연결성에 관계없이 클라우드와 같은 경험을 제공합니다.
Cerence a lancé CaLLM™ Edge, un modèle de langage embarqué de qualité automobile développé en collaboration avec Microsoft. Basé sur la famille Phi-3 de Microsoft et affiné avec le jeu de données automobile de Cerence, le modèle dispose de 3,8 milliards de paramètres et d'une taille de contexte de 4k. Il permet des commandes de contrôle du véhicule, des recherches de points d'intérêt et des interactions conversationnelles sans nécessiter de connexion Internet.
Le système fonctionne en déploiement uniquement embarqué ou hybride, garantissant une disponibilité constante des fonctionnalités d'IA générative tout en préservant la confidentialité des données en maintenant les informations à l'intérieur du véhicule. Cette solution offre aux fabricants automobiles une performance améliorée des assistants et une efficacité en coûts, tout en délivrant des expériences similaires au cloud, indépendamment de la connectivité.
Cerence hat CaLLM™ Edge eingeführt, ein eingebettetes Sprachmodell in Automobilqualität, das in Zusammenarbeit mit Microsoft entwickelt wurde. Das Modell basiert auf der Phi-3-Familie von Microsoft und wurde mit dem Automobildatensatz von Cerence feinabgestimmt. Es verfügt über 3,8 Milliarden Parameter und eine Kontextgröße von 4k. Es ermöglicht Steuerbefehle für das Auto, die Suche nach POIs und interaktive Gespräche ohne Internetverbindung.
Das System funktioniert in eingebetteten oder hybriden Einsätzen und gewährleistet eine ständige Verfügbarkeit der generativen KI-Funktionen, während es die Datensicherheit wahrt, indem Informationen im Fahrzeug bleiben. Diese Lösung bietet Automobilherstellern verbesserte Leistungsfähigkeit des Assistenten und Kosteneffizienz und ermöglicht Cloud-ähnliche Erfahrungen unabhängig von der Konnektivität.
- Development of an embedded AI system that works without internet connectivity
- Partnership with Microsoft enhances product credibility and market reach
- Cost efficiency benefits for automakers through embedded model deployment
- Integration with existing Cerence solutions portfolio
- Availability through Microsoft Azure AI model catalog expands distribution channels
- None.
Insights
This launch represents a significant technological advancement in automotive AI. The CaLLM Edge model, with 3.8 billion parameters, brings cloud-like AI capabilities directly to vehicles without requiring constant connectivity. The embedded architecture addresses key industry challenges around connectivity reliability and data privacy while maintaining cost efficiency.
The partnership with Microsoft and integration with their Phi-3 models provides strong technical validation. The 4-bit quantization and 4k context size specifications indicate an optimized balance between performance and resource usage, important for automotive applications. The ability to handle both explicit commands and conversational interactions shows promising versatility.
For automakers, this offers a competitive edge in delivering advanced AI features while controlling cloud computing costs. The hybrid deployment option provides flexibility in implementation strategies. However, success will depend on real-world performance and adoption rates by major automotive manufacturers.
CaLLM™ Edge is developed and optimized in collaboration with Microsoft, leveraging its Phi-3 family of small language models
BURLINGTON, Mass., Nov. 13, 2024 (GLOBE NEWSWIRE) -- Cerence Inc. (NASDAQ: CRNC), AI for a world in motion, today introduced CaLLM™ Edge, its pioneering, automotive-grade, embedded small language model (SLM). This new SLM is available within the existing Cerence solutions portfolio and will power its next-generation AI assistant platform, enabling an intelligent, seamless user experience regardless of connectivity. Developed and optimized in partnership with Microsoft, CaLLM Edge is available directly to Cerence’s automaker customers as well as in the Microsoft Azure AI model catalog.
CaLLM (Cerence Automotive Large Language Model) Edge is fine-tuned on Microsoft’s Phi-3 family of small language models, using Cerence’s extensive automotive dataset to deliver highly specialized AI that can handle a variety of automotive use cases. With 3.8 billion parameters, 4k context size, and 4-bit quantization, this model fits comfortably when embedded in the automotive headunit. Its core capabilities include implicit and explicit car control commands (for example, temperature, windows and doors, seat position) and point-of-interest search and navigation, as well as conversational interaction like, “What's the most popular movie ever shot in Hollywood?” followed by “Can you tell me more about the plot?”
Compatible with major automotive platforms, CaLLM Edge is available in both embedded-only deployments, meaning it can function independently without any connectivity, as well as hybrid or cloud-first deployments in which the SLM serves as one method of answering queries and as backup when connectivity is lost. For users, this means always-on access to key generative AI-style features and information, even when not connected to the cloud, as well as improved data privacy, with data staying on board in the car rather than being sent to the cloud. For automakers, CaLLM Edge delivers not only improved assistant performance, but also cost efficiency – by leveraging a fully embedded SLM model, OEMs can keep costs under control while still delivering a generative AI-based experience for their drivers.
“CaLLM Edge fundamentally transforms the way users can interact with their systems, enabling them to access the rich, responsive experiences they’ve come to expect from cloud-based systems no matter where they are,” said Nils Schanz, EVP, Product & Technology, Cerence. “As we continue to leverage our embedded and cloud expertise and advancements in generative AI and LLMs throughout our product architecture, we not only further expand the capabilities of our current solutions portfolio but also advance our next-gen platform. We’re proud to partner with Microsoft to unite our deep vertical expertise and strength in embedded solutions with their industry-leading language model capabilities.”
“Adapted AI models are becoming increasingly necessary in order to deliver intelligent, capable agent experiences across industries. We’re pleased to work with Cerence to offer CaLLM Edge in the Azure AI model catalog, giving automotive organizations the opportunity to build AI solutions in Azure AI Studio and Microsoft Copilot Studio,” said Satish Thomas, Corporate Vice President, Business & Industry Solutions, Microsoft. “Cerence’s unique automotive dataset brings incredible value, enabling a high level of accuracy and relevance throughout the user interaction – all in an embedded architecture that means the experience is available at all times.”
To learn more about Cerence, visit www.cerence.com, and follow the company on LinkedIn.
About Cerence Inc.
Cerence (NASDAQ: CRNC) is the global industry leader in creating unique, moving experiences for the mobility world. As an innovation partner to the world’s leading automakers and mobility OEMs, it is helping advance the future of connected mobility through intuitive, AI-powered interaction between humans and their vehicles, connecting consumers’ digital lives to their daily journeys no matter where they are. Cerence’s track record is built on more than 20 years of knowledge and 500 million cars shipped with Cerence technology. Whether it’s connected cars, autonomous driving, e-vehicles, or two-wheelers, Cerence is mapping the road ahead. For more information, visit www.cerence.com.
Contact Information
Kate Hickman | Tel: 339-215-4583 | Email: kate.hickman@cerence.com
FAQ
What is Cerence CaLLM Edge and how does it work?
What are the main features of Cerence (CRNC) CaLLM Edge?
How does Cerence (CRNC) CaLLM Edge benefit automakers?
What is the partnership between Cerence (CRNC) and Microsoft for CaLLM Edge?
