Скачать с ютуб 【COMPUTEX 2024 Trend Exploration Forum】Kneron: NPU Makes Personalized GPT Possible в хорошем качестве

【COMPUTEX 2024 Trend Exploration Forum】Kneron: NPU Makes Personalized GPT Possible

[English Audio by Interpreter] On May 16, the Taipei Computer Association (TCA) organized the AI PC Industry Exploration Forum as a pre-show event for COMPUTEX 2024. The event focused on the future of generative AI (GenAI) and Large Language Models (LLMs) in the tech industry. Keynote speakers from prominent companies, including Intel, Qualcomm, Google, and Kneron, shared insights on AI PC trends, enterprise and personal AI applications, and the latest developments in AI PCs. The forum also featured a panel discussion with industry leaders from Pegatron, ASUS, Quanta, LITEON, and Acer Medical. Moderated by Chris Hung from MIC, the discussion covered the positioning, functions, and practical applications of GenAI in the AI PC industry. The event attracted 700 professionals and media members, marking it as one of the most successful COMPUTEX pre-show events. Dr. Albert Liu, CEO & Founder of Kneron introduced how to use Kneron’s NPU accelerator, AI chip, and AI algorithms to design personalized GPT applications; which will need to rely on fast, accurate, and lightweight LLM models. Taking the case of edge AI computing, this can be achieved by splitting the AI model to run it on a single chip SoC architecture AI chip with built in NPU; and it will consume less power than a general GPU because users can run the GPT applications on the edge. To facilitate the operations of different Edge GPT functions, users can install or run various Edge GPT software and models the KNEO open platform to reduce development costs. Among of the main concerns for enterprises considering AI application are data confidentiality and security. To mitigate these concerns, Kneron launched the KNEO-300 which is an NPU powered edge AI server. It offers high computing speed and low energy consumption and can be applied to various enterprise GPT solutions and supports both English and Chinese languages. It has built-in GPT model training services and offline AI servers, further accelerating the deployment speed and meeting the enterprise needs for GPT computing. The entry-level version of the KNEO-300 was already launched in the end of 2023 while the enterprise version of KNEO-330 is expected to be released this year, which will allow SMEs to create their own private GPT applications. 【COMPUTEX Official Website】 🌐COMPUTEX Website: http://www.computex.biz/ 🌐COMPUTEX CYBERWORLD: https://show.computex.biz/online.aspx 🌐COMPUTEX Best Choice Award: https://bcaward.computex.biz/ 🌐COMPUTEX Facebook:   / computextaipei   🌐COMPUTEX Instagram:   / computextaipei   🌐COMPUTEX LinkedIn:   / computex-taipei   #computex #computex2024 #ai #aipc