TAIPEI (TVBS News) — Hon Hai Research Institute (HHRI, 鴻海研究院) on Monday (March 10) unveiled its first traditional Chinese large language model (LLM), FoxBrain, showcasing exceptional computational efficiency.
The model, initially designed for internal use, promises to revolutionize data analysis, decision support, and document collaboration, among other functions. Hon Hai plans to open-source FoxBrain, aiming to enhance its application beyond internal operations.
Director Li Yong-hui (栗永徽) of the Artificial Intelligence Research Institute at HHRI emphasized the strategic focus on optimizing training processes. The model underwent training using 120 NVIDIA H100 GPUs, leveraging NVIDIA Quantum-2 InfiniBand networks, which was completed in just four weeks.
Hon Hai aims to integrate FoxBrain into its smart manufacturing, smart electric vehicle, and smart city platforms, enhancing data analysis efficiency. This move positions FoxBrain as a crucial engine driving the upgrade of intelligent applications, potentially boosting the company's operational benefits.