A breakthrough has been achieved by a team of researchers from Tokyo Institute of Technology, in collaboration with Fujitsu and other partners, who have successfully developed a large language model on the Japanese supercomputer Fugaku. This innovative model will lay the groundwork for generative artificial intelligence.
The Fugaku-LLM model has been extensively trained on Japanese language data, which accounts for a majority of the total training data. The unveiling of this model is set to pave the way for further research on generative AI customized to meet domestic requirements.
Since May 2023, the researchers from Tohoku University, Nagoya University, Riken, CyberAgent, and Kotoba Technologies have been working together on this project utilizing the cutting-edge supercomputer jointly developed by Fujitsu and Riken.