The mannequin, named “FoxBrain,” was skilled utilizing 120 of Nvidia’s H100 GPUs and accomplished in about 4 weeks, the world’s largest contract electronics producer mentioned in an announcement.
The firm, which assembles iPhones for Apple and in addition produces Nvidia’s synthetic intelligence servers, mentioned the mannequin is predicated on Meta’s Llama 3.1 structure.
It is Taiwan’s first giant language mannequin with reasoning capabilities that’s optimised for conventional Chinese and Taiwanese language kinds, it mentioned.
Foxconn mentioned that although there was a slight efficiency hole in contrast with China’s DeepSeek’s distillation mannequin, its total efficiency could be very near world-class requirements.
Initially designed for inside purposes, FoxBrain covers knowledge evaluation, choice help, doc collaboration, arithmetic, reasoning and problem-solving, and code era.
Foxconn mentioned it plans to collaborate with know-how companions to increase the mannequin’s purposes, share its open-source info, and promote AI in manufacturing, provide chain administration, and clever decision-making.
Nvidia offered help by means of its Taiwan-based supercomputer “Taipei-1” and provided technical consulting in the course of the mannequin’s coaching, Foxconn mentioned.
Taipei-1, the biggest supercomputer in Taiwan, is owned and operated by Nvidia in Kaohsiung, a southern metropolis on the island.
Foxconn will announce additional particulars in regards to the mannequin throughout Nvidia’s GTC developer convention in mid-March.
© Thomson Reuters 2025
(This story has not been edited by NDTV workers and is auto-generated from a syndicated feed.)
