The primary objective of AutoLang is to adapt generalist, open-source LLMs to the automotive domain to help assist/automate language-based tasks in automotive engineering. Many task artifacts and work products of automotive engineering are natural language documents. The corollary, of course, is that automotive engineers spend significant effort creating/reviewing these documents, or even taking actions based upon compiled documents. An AutoLang family of LLMs adapted to documents from automotive domains has the potential to assist/automate document-based engineering, significantly improving productivity.
Looking ahead, it’s conceivable to apply this approach to other domains such as telecommunications or healthcare, with much of the underlying logic remaining applicable. This project sets a strong foundation for future research into domain-specific large language models.