March 10 (Bloomberg) -- According to Reuters.Hon Hai Precision Industry Company, Taiwan technology companytoday announced the launch of the firstLarge Language Model“FoxBrain" and plans to use the technology to optimize manufacturing and supply chain management.
Hon Hai said in a statement that FoxBrain is operated by the 120 NVIDIA H100 GPUs The training is completed with a training cycle of about four weeks. Hon Hai is now an Apple iPhone assembler and maker of NVIDIA's AI servers, and is the world's largest electronics foundry.
The model is based on Meta's Llama 3.1 architecture and has been specifically optimized toAdaptation of Traditional Chinese and local language styles. Honghai says it is the first large-scale language model with reasoning capabilities in the region. It claims that although FoxBrain is performanceSlightly inferior to DeepSeek's distillation modelbut the overall performance has beenClose to the top of the world.
FoxBrain is primarily used for internal scenarios and supports theData Analysis, Decision Aids, Document Collaboration, Math Operations, Reasoning and Problem Solving, and Code Generation.
Hon Hai plans to work with technology companies to expand the model's application, whileopen sourcethat drives artificial intelligence in manufacturing, supply chain management and intelligent decision-making.
NVIDIA also supports FoxBrain training by providing computing power through its supercomputer "Taipei-1" in Kaohsiung, as well as technical guidance during the training process.
Note: "Taipei-1" is the largest supercomputer in the region, operated by NVIDIA in Kaohsiung.