Chip Maker Qualcomm Unveils AI Hub – Announces Key Advancements

The Qualcomm AI Hub serves as a gateway for developers seeking superior on-device AI performance. This library boasts over 75 pre-optimized AI models, including popular ones like Whisper, ControlNet, Stable Diffusion, and Baichuan 7B.
Chip Maker Qualcomm Unveils AI Hub – Announces Key Advancements
Image Source: Qualcomm

Major chip maker Qualcomm recently announced a suite of new AI-related initiatives at the Mobile World Congress (MWC) in Barcelona. It includes the Qualcomm AI Hub, research breakthroughs, and a display of AI-enabled devices, among a diverse range of devices, powered by Snapdragon and Qualcomm platforms.

Key highlights:

  • The Qualcomm AI Hub serves as a gateway for developers seeking superior on-device AI performance. This library boasts over 75 pre-optimized AI models, including popular ones like Whisper, ControlNet, Stable Diffusion, and Baichuan 7B. 

  • Developers can integrate these models into their applications, leveraging the Qualcomm AI Engine to get four times faster inferencing times. These models are tailored for improved on-device AI performance, reduced memory use, and improved power efficiency across various form factors.

"We are thrilled to host Qualcomm Technologies' AI models on Hugging Face," said Clement Delangue, Co-founder and CEO of Hugging Face. "These popular AI models optimized for on-device machine learning and ready to use on Snapdragon and Qualcomm platforms will enable the next generation of mobile developers and edge AI applications, making AI more accessible and affordable for everyone."

  • The models are readily available on the Qualcomm AI Hub, GitHub, and Hugging Face. The Hub's AI model library claims to facilitate quicker integration, resulting in faster time-to-market, and tap into the advantages of on-device AI, such as immediacy, reliability, privacy, personalization, and cost savings.

  • Qualcomm AI Research is unveiling advancements, including the Large Language and Vision Assistant (LLaVA), an over seven billion parameter LLM. LLaVA operates on Android smartphones, accepting various data inputs like text and images to generate multi-turn conversations with an AI assistant about an image.

  • Another breakthrough is the demonstration of Low Rank Adaptation (LoRA) on Android smartphones, enabling users to create high-quality custom images based on personal or artistic preferences. Qualcomm AI Research is also showcasing a 7+ billion parameter LMM on a Windows PC that can generate multi-turn conversations about audio inputs. These models prioritize privacy, reliability, personalization, and cost efficiency.

  • Smartphones: Flagship commercial AI smartphones, including HONOR Magic6 Pro, OPPO X7 Ultra, and Xiaomi 14 Pro, feature generative AI capabilities such as AI-generated image expansion, AI-powered video creation, and AI-powered calendar creation.

  • PCs: The Snapdragon X Elite, equipped with a 45 TOPS NPU, transforms PC interactions with on-device AI. Qualcomm demonstrates the efficiency of generative AI in image creation, outpacing x86 competitors.

  • Automotive: Utilizing AI hardware and software solutions, Qualcomm demonstrates traditional and generative AI capabilities with the Snapdragon® Digital Chassis™ Platform, promising powerful, efficient, private, safer, and personalized experiences for drivers and passengers.

  • Consumer IoT: The showcase includes Humane's AI Pin, running on a Snapdragon platform, offering users an innovative, conversational, and screenless AI experience.

  • Connectivity: Qualcomm introduces the Snapdragon X80 Modem-RF System, integrating a second-generation 5G AI processor, and the Qualcomm FastConnect 7900 Mobile Connectivity System, the first AI-optimized Wi-Fi 7 system.

  • 5G Infrastructure: Qualcomm unveils three AI-based enhancements for network management, simplifying tasks for radio access network engineers, reducing network energy consumption, and optimizing 5G network slice lifecycle management.

Related Stories

No stories found.
CDO Magazine
www.cdomagazine.tech