Baobab has achieved the highest accuracy validation result in Japan, second only to Llama2, in the development of a domestically produced purpose-specific Japanese LLM (large-scale language model) in collaboration with Ridge-i Inc.

 

The R&D team included Baobab, Ridge-i Corporation, and Graham Neubig, Associate Professor at Carnegie Mellon University.

 

Reference: Participation in Large-scale Language Model (LLM) R&D in Japan (July 19, 2023).

 

Please refer to Ridge-i’s announcement for details.

https://prtimes.jp/main/html/rd/p/000000072.000026963.html

 

The results of this validation confirm once again the importance of task-specific data.

Baobab will continue to meet the ever-increasing need for fine tuning with the most experience and the most complete system in the country.