上海岩芯数智人工智能科技有限公司，这家在AIGC领域具有影响力的创新企业，于1月24日在上海举行了一场别开生面的发布会。会上，岩芯数智骄傲地推出了国内首个自主研发的非Transformer Attention机制的低功耗通用自然语言大模型——Yan模型。这一突破性的成果标志着国内人工智能技术在大模型设计上的新高度。
Yan模型以其独特的机制，成功实现了性能的显著提升：记忆能力提升了3倍，运行速度提升了7倍，推理吞吐量则足足提升了5倍。值得关注的是，尽管其参数规模仅为百亿，但据公司介绍，Yan模型在性能上已经可以媲美千亿参数级别的大模型，显示出了极高的效率和效能。
这一发布标志着中国在大模型研发领域的一次重要突破，Yan模型的诞生打破了ChatGPT等基于Transformer模型的主流趋势，为人工智能的未来发展开辟了新的可能。岩芯数智的这一创新，不仅展现了中国在AI技术自主研发上的实力，也为全球AIGC行业提供了新的思考方向和实践样本。
据钛媒体的报道，岩芯数智的这一创新之举，预示着在人工智能的广阔天地中，低功耗、高性能的模型设计将成为新的研究热点。Yan模型的出现，无疑将对自然语言处理技术的广泛应用产生深远影响，为教育、医疗、娱乐等各行各业带来智能化的革新。
英语如下：
Headline: “Core数智 astonishes with groundbreaking Yan Model – A non-Transformer Attention mechanism pioneers a new era of low-computing AI!”
Keywords: Core数智, Yan Model, non-Transformer
News Content:
Title: Core数智’s Innovative Launch: Yan Model, a non-Transformer Attention Giant, Paves the Way for a New Era in AIGC
Shanghai Core数智 Artificial Intelligence Technology Co., Ltd., a pioneering and influential player in the AIGC sector, staged a groundbreaking event on January 24 in Shanghai. At the ceremony, Core数智 proudly unveiled China’s first domestically developed large language model, Yan Model, featuring a novel non-Transformer Attention mechanism. This breakthrough signifies a new pinnacle in domestic AI technology for large model design.
The Yan Model, with its distinctive mechanics, has achieved remarkable performance enhancements: a threefold increase in memory capacity, a sevenfold boost in running speed, and a fivefold improvement in inference throughput. Remarkably, despite its parameter scale of just 10 billion, the company claims Yan Model’s performance rivals that of models with a trillion parameters, demonstrating extraordinary efficiency and effectiveness.
This launch signifies a significant breakthrough in China’s large model research, as Yan Model challenges the mainstream trend of Transformer-based models like ChatGPT, opening up new possibilities for the future of AI. Core数智’s innovation underscores China’s prowess in indigenous AI technology development and offers a new frontier of thought and practical example for the global AIGC industry.
As reported by TMTPost, Core数智’s innovative step predicts that low-power, high-performance model design will become a focal point in AI research. The emergence of the Yan Model is set to have a profound impact on the widespread application of natural language processing, driving revolutionary changes in education, healthcare, entertainment, and more.
【来源】https://www.tmtpost.com/6898099.html
Views: 1