Reusing knowledge from a pretrained model and adapting it to a new task or domain via fine‑tuning or prompting strategies.