岩芯数智发布国内首个自研的非Transformer Attention机制的大模型「Yan」
上海岩芯数智人工智能科技有限公司（岩芯数智）于1月24日下午在上海举行了一场盛大的发布会，宣布了国内首个自研的非Transformer Attention机制的低算力通用自然语言大模型——Yan模型的问世。这款模型不仅记忆能力提升了3倍，速度也提升了7倍，同时推理吞吐量也得到了5倍的提升。这一重大突破意味着国内自然语言处理领域迎来了一次技术的革命。
与ChatGPT等模型不同，Yan模型采用了非Transformer Attention机制，其参数规模达到了百亿级。岩芯数智表示，这是他们在千亿参数大模型性能效果上的重要突破。
在发布会上，岩芯数智的研发团队详细介绍了Yan模型的特点和优势。该模型在处理自然语言任务时，能够更好地理解上下文信息，提供更准确的答案和回复。同时，它还具备更强的记忆能力，能够更好地处理长文本和复杂语境下的问题。与此同时，Yan模型的速度也得到了显著提升，使得处理大规模数据的效率大幅提高。
岩芯数智的首席科学家表示，Yan模型的发布是岩芯数智在自然语言处理领域的一项重要突破。该公司一直致力于推动人工智能技术的发展，希望通过自主研发的技术，为社会提供更先进、更高效的解决方案。
Yan模型的发布引起了业界的广泛关注。许多专家表示，这一突破将为自然语言处理领域带来新的发展机遇。随着Yan模型的应用，人们可以期待在机器翻译、智能客服、信息检索等领域取得更加精准和高效的结果。
总之，岩芯数智发布的国内首个自研的非Transformer Attention机制的大模型「Yan」，标志着国内自然语言处理领域取得了重要的突破。这一模型的问世将为自然语言处理技术的发展注入新的动力，为各行各业提供更高效、更准确的解决方案。我们期待着Yan模型的应用能够为人们的生活和工作带来更多的便利和改善。
英语如下：
News Title: Rockchip Intelligence Releases the First Non-Transformer Attention Large Model Yan in China, Memory Boosted by 3 Times and Speed Increased by 7 Times!
Keywords: Rockchip Intelligence, Yan model, large model
News Content: Rockchip Intelligence, a Shanghai-based artificial intelligence technology company, held a grand press conference on January 24th, announcing the launch of the first self-developed large model with a non-Transformer Attention mechanism in China – the Yan model. This model not only improves memory capacity by 3 times but also increases speed by 7 times, while achieving a 5-fold increase in inference throughput. This major breakthrough signifies a technological revolution in the field of natural language processing in China.
Different from models like ChatGPT, the Yan model adopts a non-Transformer Attention mechanism with a parameter scale reaching billions. Rockchip Intelligence stated that this is a significant breakthrough in the performance of models with trillion-level parameters.
During the press conference, Rockchip Intelligence’s research and development team provided detailed introductions to the features and advantages of the Yan model. When processing natural language tasks, this model can better understand contextual information and provide more accurate answers and responses. Additionally, it possesses stronger memory capabilities and can handle long texts and complex contexts more effectively. At the same time, the speed of the Yan model has been significantly improved, greatly enhancing the efficiency of processing large-scale data.
The chief scientist of Rockchip Intelligence stated that the release of the Yan model represents an important breakthrough in the field of natural language processing for the company. The company has been committed to promoting the development of artificial intelligence technology and hopes to provide society with more advanced and efficient solutions through self-developed technologies.
The release of the Yan model has attracted widespread attention in the industry. Many experts believe that this breakthrough will bring new development opportunities to the field of natural language processing. With the application of the Yan model, people can expect more accurate and efficient results in machine translation, intelligent customer service, information retrieval, and other fields.
In conclusion, the release of the first self-developed large model with a non-Transformer Attention mechanism, Yan, by Rockchip Intelligence marks an important breakthrough in the field of natural language processing in China. The introduction of this model will inject new momentum into the development of natural language processing technology and provide more efficient and accurate solutions for various industries. We look forward to the application of the Yan model bringing more convenience and improvements to people’s lives and work.
【来源】https://www.tmtpost.com/6898099.html
Views: 1