You are working on a system that needs to understand clinical notes, including shorthand, section structure, medication mentions, diagnoses, and assessment language. A general language model is not reliable enough on this kind of text, so you want to adapt it to the domain.
How would you fine tune an LLM for clinical note understanding?