Samsung Gauss2 Multimodal AI Model Supporting Up to 14 Languages Presented at SDC24
Samsung on Thursday unveiled the second generation of its Gauss artificial intelligence (AI) model. The new multimodal AI model, called Gauss2, is said to feature improved performance and efficiency, as well as improved use cases for application integration. The new Large Language Model (LLM) was unveiled during a keynote at the Samsung Developer Convention Korea 2024 (SDC24 Korea), which was held online. The tech giant emphasized that it is using the power of the artificial intelligence model for software research and development purposes.
Samsung Gauss2 AI model unveiled
In a news release, the South Korean tech giant detailed the second generation of its Gauss AI model released last year. The new version of the base model now includes several updates. The company said that multimodal capabilities have been improved, which include processing datasets in multiple modalities.
Additionally, the AI model now supports 9 to 14 languages, as well as multiple programming languages. It is also said to improve the performance of language, code, and images.
Samsung Gauss2 is available in three different models, depending on the size of its parameters: Compact, Balanced and Supreme. The Compact is a small-size model designed for efficiency and performance in limited computing environments. The Balanced model is optimized for both performance and efficiency, while the Supreme model can solve high-level computational problems by leveraging MoE technology.
Samsung says the Balanced and Supreme models can outperform “leading open-source generative AI models” on English, Korean, and coding tasks. The company also claims that these models provide reduced latency, faster processing speeds, and better task completion compared to open-source models.
Samsung Gauss2 is currently being used by the company’s System eXperience (DX) division and overseas research institutes. The most popular use case for an AI model is an internal coding assistant called code.i, which helps in software development. The tech giant said 60 percent of its DX division uses the tool. The company’s call center employees also use the technology to categorize and summarize customer calls.
Although Gauss2 is currently used internally, the company also plans to ship it with its products. Samsung also believes that the AI model can improve the customization capabilities of existing AI features.
Five indicted in US for hacking cryptocurrency using ‘scatter spider’ method: details