Incremental fine-tuning: any way or workaround?

Hi all, do you know of any way to incremental fine-tune?
I’ve found nothing about it.

I’m working on a repository, so that the AI gets the data and the information from the papers I fine-tune.

Because I work with academic papers, I need my AI to be always updated, so that I can add a new piece of knowledge when a new research in my field goes out.