Qualcomm Partners with Tencent's HunYuan for Enhanced On-Device AI Applications: Snapdragon 8 Gen 2 for Galaxy Enables Generative AI Adoption

Qualcomm Partners with Tencent's HunYuan for Enhanced On-Device AI Applications: Snapdragon 8 Gen 2 for Galaxy Enables Generative AI AdoptionAt the 2024 Snapdragon Summit, Qualcomm announced a partnership with Tencent's HunYuan large language model. Together, leveraging the Snapdragon 8 Gen 2 for Galaxy mobile platform, they successfully deployed the 7B and 3B versions of Tencent's HunYuan on-device, showcasing impressive performance results

Qualcomm Partners with Tencent's HunYuan for Enhanced On-Device AI Applications: Snapdragon 8 Gen 2 for Galaxy Enables Generative AI Adoption

At the 2024 Snapdragon Summit, Qualcomm announced a partnership with Tencent's HunYuan large language model. Together, leveraging the Snapdragon 8 Gen 2 for Galaxy mobile platform, they successfully deployed the 7B and 3B versions of Tencent's HunYuan on-device, showcasing impressive performance results. This collaboration will empower Tencent's HunYuan to provide robust technical support for a wide range of business scenarios, accelerating product innovation through on-device AI, effectively reducing operating costs, and further expanding the adoption and deployment of generative AI on devices.

The Snapdragon 8 Gen 2 for Galaxy mobile platform is equipped with the new second-generation custom Qualcomm Oryon CPU and enhanced Qualcomm Hexagon NPU, leveraging the advantages of the Qualcomm AI Engine to deliver powerful on-device generative AI processing capabilities. The Snapdragon 8 Gen 2 for Galaxy's robust AI computing power, combined with Qualcomm's AI software stack and industry-leading toolkits, including the Qualcomm AI Model Efficiency Toolkit (AIMET), provides comprehensive optimization capabilities for the model. Using hardware-based INT4 quantization technology significantly enhances the operational efficiency of Tencent's HunYuan on devices, achieving a first token generation latency of 150ms and a decoding rate exceeding 30 tokens per second on device inference.

Tencent's HunYuan large language model already provides underlying technical support for over 700 internal business scenarios and C-end applications at Tencent, including WeChat input method, Tencent Mobile Manager, QQ, Tencent Video, QQ Browser, Enterprise WeChat, Tencent Meeting, etc. By achieving on-device deployment targeted at the Snapdragon 8 Gen 2 for Galaxy, it can leverage the rich advantages of on-device generative AI to better meet the diverse business needs of devices.

For example, Tencent Mobile Manager's intelligent SMS identification feature leverages the on-device model capabilities of Tencent's HunYuan for the first time. Combining massive data with deep neural networks and pre-training, the model exhibits exceptionally strong semantic understanding capabilities. By integrating contextual information, it more accurately understands the intent of SMS messages, resulting in a near 200% increase in SMS recall rate and a 20% improvement in identification accuracy. As some SMS messages involve sensitive personal information of users, on-device AI can effectively safeguard user privacy while maintaining excellent performance.

"Qualcomm and Tencent's HunYuan team have been working together for a long time to drive the deployment and promotion of cutting-edge on-device technology innovations in mobile applications, delivering innovative application experiences for users and consumers," said Durga Malladi, Senior Vice President and General Manager, Technology Planning and Edge Solutions, Qualcomm Technologies, Inc. Qualcomm is committed to empowering ISV partners and developers to leverage Qualcomm's powerful heterogeneous computing, industry-leading CPUs, GPUs, NPUs, and software solutions to drive adoption of generative AI applications on Snapdragon-powered devices, benefiting more users worldwide."

"Tencent's HunYuan has established a full-link large model matrix and application platform. At the same time, Tencent HunYuan is constantly perfecting its deployment ecosystem. In cloud service scenarios, we are continuously reducing call prices while improving effectiveness through various technologies. In intelligent terminal scenarios, Tencent and Qualcomm continue to deepen their collaboration, empowering different business needs for on-device model deployment and continuously iterating capabilities within Tencent's extensive ecosystem, allowing more B-end and C-end users to enjoy the extraordinary experience of practical large models," said Wang Di, General Manager of Tencent's Machine Learning Platform Department and Head of Tencent's HunYuan large language model.

The success of this collaboration not only marks significant progress for Qualcomm and Tencent in driving on-device AI applications but also further promotes the adoption and application of generative AI technology. In the future, both parties will continue to deepen their collaboration to jointly explore new breakthroughs in on-device AI technology, bringing users a more intelligent, efficient, and convenient mobile application experience.

The Snapdragon 8 Gen 2 for Galaxy's powerful AI computing power and Qualcomm's AI software stack provide a solid foundation for Tencent's HunYuan large language model deployment on devices.

Tencent's HunYuan large language model's semantic understanding capabilities and extensive application scenario coverage offer rich application cases for the popularization of on-device generative AI applications.

The collaboration between Qualcomm and Tencent will accelerate the development and deployment of generative AI technology on devices, bringing users a more intelligent mobile experience.

Disclaimer: The content of this article is sourced from the internet. The copyright of the text, images, and other materials belongs to the original author. The platform reprints the materials for the purpose of conveying more information. The content of the article is for reference and learning only, and should not be used for commercial purposes. If it infringes on your legitimate rights and interests, please contact us promptly and we will handle it as soon as possible! We respect copyright and are committed to protecting it. Thank you for sharing.(Email:[email protected])