Fine tuning an LLM on a 2M word diary

Interesting. why didn’t you train on gpt2-do or a more recent model out of interest?

(and accidentally curing my memory hoarding ocd)