The project, which began in May 2023, involved researchers from several other institutions such as Tohoku University, Nagoya University, the government-backed research institute Riken, CyberAgent, and Kotoba Technologies. The supercomputer used in the project was a joint development by Fujitsu and Riken.
Key takeaways:
- A team of researchers from various institutions, including the Tokyo Institute of Technology and Fujitsu, have developed a large language model for generative artificial intelligence using the Japanese supercomputer Fugaku.
- The model, known as Fugaku-LLM, was trained extensively on Japanese data, which made up 60% of the total training data.
- The Fugaku-LLM model is expected to lead to research on generative AI tailored to domestic needs in Japan.
- The project was launched in May 2023 and involved researchers from Tohoku University, Nagoya University, Riken, CyberAgent, and Kotoba Technologies, in addition to the supercomputer jointly developed by Fujitsu and Riken.