News for hackers. Stuff that matters.
Re:Developing an LLM: Building, Training, Finetuning (A 1h V... by rasbt on Friday June 14, 2024 @09:14AM (#40680622)
I wouldn't pretrain from scratch, but continued pretraining is pretty popular for adapating LLMs to recent and/or custom data. (Sometimes this is referred to 'finetuning', however, not to be confused with 'instruction finetuning').

There are no comments on this comment yet.