Foxconn Launches First Chinese LLM to Advance AI in Manufacturing

Foxconn Technology Group, the largest electronics contract manufacturer globally and key iPhone supplier for Apple, introduced its firsrt Chinese large language model (LLM) trained on traditional characters, as the Taiwanese firm advances the implementation of artificial intelligence (AI) in its factories.

The updated FoxBrain model was developed in a “more efficient and lower-cost " way in only four weeks, establishing a new benchmark in Taiwan's AI technology advancement, as stated on Monday by Foxconn, officially known as Hon Hai Precision Industry.

According to Foxconn, FoxBrain demonstrates exceptional skills in mathematics and logical reasoning with a training process powered by 120 Nvidia H100 graphics processing units (GPUs).

Initially intended for internal use within the company, Foxconn announced plans to make it open source in the future to foster collaboration with technology partners, broaden its applications, and advance AI in manufacturing.

LLMs are the technology that supports generative AI services such as OpenAI's ChatGPT. Open source provides public access to a software's source code, enabling third-party developers to alter or distribute its design, repair broken links, or enhance its functionalities.

The latest model utilized the Meta Llama 3.1 framework, featuring 70 billion parameters. Foxconn claimed that it surpassed Llama-3-Taiwan-70B, an additional open-source model refined with traditional Chinese characters and English data utilizing the Llama-3 framework, in the majority of categories of TMMLU+, a standard for traditional Chinese language understanding.