LLM Research and Development Initiatives at the National Institute of Informatics

Explore the comprehensive research and development initiatives for large language models at Japan's National Institute of Informatics in this 18-minute conference presentation. Learn about the establishment of the Large Language Model Research and Development Center and the ongoing llm-jp (LLM Study Group) activities aimed at building high-performance LLMs optimized for Japanese language tasks with enhanced transparency and reliability. Discover specific examples of released models, including detailed approaches to corpus construction for training, training models from scratch, fine-tuning methodologies, evaluation frameworks, and multimodal capabilities development. Gain insights into how academic institutions are advancing LLM research with a focus on language-specific optimization and responsible AI development practices.