News for hackers. Stuff that matters.

A comment on Show HN: How we got fine-tuning Mistral-7B to not suck. Parent comment

Re:Show HN: How we got fine-tuning Mistral-7B to not suck by gpderetta on Thursday February 08, 2024 @06:18AM (#39300661)
> If you have an infinite amount of facts, then you can memorise all of them

pigeon-hole?

[ Parent | on HN ]
Re:Show HN: How we got fine-tuning Mistral-7B to not suck by gdiamos on Thursday February 08, 2024 @07:30AM (#39301147)
Not literally infinite, but Llama2 scale models can handle about 10 trillion tokens.
[ Parent | on HN ]