Model behind Translation LLM and Adaptive Translation

Can someone clarify what model does Translation LLM and Adaptive translation uses? If there’s some source which validates it, it would be great.